Add scripts to upload the build results. They were taken from the
dora branch and got the directory names adjusted. The rest of the
script should work okay.
Add explicit reference to osmo-gbproxy.service to fix build issue.
ERROR: QA Issue: openbsc: Files/directories were installed but not shipped
/lib/systemd/system/osmo-gbproxy.service
ERROR: QA run found fatal errors. Please consider fixing them.
The old Geode based systems are replaced by AMD APU systems. We might
build for sysmobsc v1 and v2 but would like to be able to tune for the
better CPU.
This is fixing the QA issue on newer perf for x86 on Dora as seen
by the CI system. Drop support for ancient unsupported OE version
and force 3.10/4.9 depending on where we are.
mtd-utils-native needs to be available for do_deploy, so add the
dependency there. DEPENDS is meaningless for us so remove it too.
Use specific tasks as dependency to workaround an issue with poky
master where glibc-initial and glibc-locale would be direct deps.
Clone the recipe to deal with the different tasks. I wasn't able
to get the other tasks in place for dora.
In modern Poky the facade is gone and the data dictionary needs
to be accessed directly. This commit is catching up with it. It
was build tested on dora as well. I have not checked if it is
changing the checksums of a dora build.
Take a patch from newer version in Poky to hardcode the os to
Linux. This surfaced because of the more strict sysroot in newer
versions of Poky.
To avoid some bad surprises I am continuing to use this version
of ifupdown instead of the newer one in Poky.
The function is declared and defined as inline and gcc6 will not
emit an external visible symbol for it. Master doesn't have this
problem anymore so let's just patch it here.
Fixes:
| /usr/src/debug/osmo-bts/0.4.3+gitAUTOINC+57ee780789-r1a2ff06d20-r18.18/git/tests/sysmobts/sysmobts_test.c:207: undefined reference to `l1if_ms_pwr_ctrl'
In newer Poky the LDFLAGS are not passed through the $CC variable
anymore and we need to make sure scons is picking them up. Take an
old Gentoo patch for it.
In commit a7e967225d we added
python-argparse to the SDK/toolchain. However, upstream dora doesn't
have this in the nativesdk bbclass, so let's add it using a .bbappend
file. I tried with python-argparse_%.bbappend to be less version
dependent, but that didn't work.
Only sysmobts-v2 will override it, but all other machines should have
the same configuration as the sysmocom-bsc. This will fix the build of
our yocto-dora for sysmocom-idu/sysmocom-odu, where task-sysmocom-feed
pulls in task-gprscore which in turn wants to build the ggsn
configuration even on those machines where we don't actually have a ggsn
so far.
Yocto master has switched to use python3 and this means all of
our bbclasses must work with it. The cmp function has been removed
and the closest is this dropin routine. Start to use it.
which (at least for owhw-v2) fixes the automatic discovery of the
bootstate backend EEPROM. I suspect it should now also work without
the hardc-de-layout-values.patch on sysmobts-v2 and sysmocom-odu, but
that remains to be tested (and thus the patches remain).
UBL is setting up the flash timings (A1CR) and so far U-Boot
was overwriting it with a timing for the EVB development board
of TI. Let UBL define ECC/ChipSelect and the A1CR timings.
This might now cause different kind of flash issues because the
per chip timings might be "wrong". This is why this commit should
needs to stay in testing a bit longer. We should use this commit
on RevC, D, E and F hardware ourselves a bit.
I didn't know that u-boot was creating a package and installing
the bin into /boot. The detection of the file seems to have
changed and a new QA error comes up.
ERROR: u-boot-v2011.12+gitAUTOINC+f6453a3911-r0 do_package: QA Issue: split_and_strip_files: 'file /home/oebuilds/jenkins/workspace/Yocto-Master/label/OE/build/tmp/work/sysmobts_v2-poky-linux-gnueabi/u-boot/v2011.12+gitAUTOINC+f6453a3911-r0/package/boot/u-boot-sysmobts-v2-v2011.12+gitAUTOINC+f6453a3911-r0' failed [split-strip]
This seems to have been introduced by changing the LC_ALL to
en_US.UTF-8. I have started a dialogue with upstream development
but as we do not need a u-boot package, let's just remove the
files.
ERROR: image-rauc-ubi-1.0-r0 do_populate_lic: QA Issue: image-rauc-ubi: Recipe file does not have license file information (LIC_FILES_CHKSUM) [license-checksum]
Exclude DATETIME from the variables to fix errors like:
ERROR: meta-toolchain-osmo-1.0-r7 do_populate_sdk: Taskhash mismatch 0fb3116488402f87d1e4d637af229664 verses 6dda5ee949e484709fbac4a097018fc2 for meta-telephony/recipes-osmocom/meta/meta-toolchain-osmo.bb.do_populate_sdk
This glib-2.0 version was required by rauc and should have been
placed in the recipes-extra (or recipes-fixes) directory.
Yocto master has imported gobject-introspecition and this requires
a newer version of glib-2.0. For Yocto master this recipe has
replaced the newer one. By moving it to the yocto-dora directory
we continue to use the old version on dora and the default version
on master.
This will allow us to have multiple versions installed at the
same time. Use a HACK to not have package.bbclass renamed the
RREPLACE we need for the upgrade.
Fixes: SYS#217
Disabling the new "binfmt" doesn't work and breaks the build. I have
sent an email to the mailinglist and this might be fixed but at the
same time binfmt is split into a new package anyway. At the same time
we should enable some items we have now switched off. Once we start
to use/stabilize this build we need to check what we need and what the
cost is.
It appears that in latest poky master the CC variable is not automatically
passed to the build. The last good build was 7cd835177a
and now it was failing. Pass CC to the build.
Yocto is now using iproute2 4.4 and the tools we packaged into
the main package are now available as extra packages and we can
just drop our extension for master.
libexecdir is now a dedicated directory and empty in our ntpd
build. Avoid QA issue of an empty directory not belonging to
anyone. Clean it if it is empty
Symptoms with LCR: nta outgoing create: invalid URI
Take patch posted to the upstream project and carried by Debian
and Ubuntu for this project. Unroll the different fields by hand
to fix undefined behavior.
Exception: OSError: [Errno 36] File name too long: '/home/oebuilds/jenkins/workspace/Yocto-Master/label/OE/build/tmp/deploy/sources/allarch-poky-linux/Firmware-AbilisFirmware-agereFirmware-amd-ucodeFirmware-atheros_firmwareFirmware-broadcom_bcm43xxFirmware-ca0132Firmware-chelsio_firmwareFirmware-cw1200Firmware-dib0700Firmware-ene_firmwareFirmware-fw_sst_0f28Firmware-go7007Firmware-i2400mFirmware-ibt_firmwareFirmware-it913xFirmware-iwlwifi_firmwareFirmware-IntcSST2Firmware-MarvellFirmware-mwl8335Firmware-myri10ge_firmwareFirmware-OLPCFirmware-phanfwFirmware-qat_dh895xcc_firmwareFirmware-qla2xxxFirmware-r8a779x_usb3Firmware-radeonFirmware-ralink_a_mediatek_company_firmwareFirmware-ralink-firmwareFirmware-rtlwifi_firmwareFirmware-tda7706-firmwareFirmware-ti-connectivityFirmware-ueagle-atm4-firmwareFirmware-via_vt6656Firmware-wl1251Firmware-xc4000Firmware-xc5000Firmware-xc5000cFirmware-sianoFirmware-qualcommAthos_ar3kFirmware-qualcommAthos_ath10k'
Our initramfs images are supposed to be small and don't have a
/etc/opkg folder so attempting to put feed config in there will
fail. Reset the FEED_URIS that come from our local.conf.
When restoring the backup the rootfs is mounted in /rootfs/data
and the symlink to the real file would then not work. Attempt to
change the directory first to get a relative symlink.
Untested change and needed for rauc
In dora we do not have autotools-brokensep. Do it the old way
and set the B variable directly. At sysmocom we still use this
layer for the very old Dora release.
* libpcap is no longer a dependence.
* libusb is needed for building fpga_load, needed for firmware loading
of the Xorcom Astribank.
* do_configure, do_compile and do_install were removed to use functions
provided by autotools class.
Signed-off-by: Fabio Berton <fabio.berton@ossystems.com.br>
The configuration scripts are named for the systemd units for which they
generate configuration files. The generator causes them to run before
those units are started.
* the ntpdate cron job turned into
a shell script which now checks
for running timeserver before
updating the time via ntpdate
* install the cron stanza to the
cron.d directory like serval
other packages does
* comment the entry in the old
crontab file when found
Fixes: SYS#1841
Yocto master has introduced GCC5 and this doesn't like to compile
the older linux kernels yet. This year it is time to move to a newer
LTS kernel and this can be Linux-4.1. Make space for the upgrade
by moving the recipe to a new name.
This change adds an initramfs module to mount the data partition and
restore the backup to the overlay. It is enabled by adding it to the
image-rauc-slot-initramfs recipe.
Signed-off-by: Jan Luebbe <jluebbe@debian.org>
[hfreyther: keep udev out of the dependencies]
The new version searches mounted slots for matching files referenced by
the new manifest. This avoids downloading files we already have on the
system.
Additionally, the new version includes better error handling and more
unit tests.
Signed-off-by: Jan Luebbe <jluebbe@debian.org>
Journald being the typical GNOME application is leaking memory left
and right. Let's backport some of the patches that were done in
later versions.
Fixes: SYS#1843
* use printf instead of echo
* add missing /etc/ for the udhcpc-default file
* add log-entries for all shell actions in the
post image setup
* generate comments in all files which are touched
Related: SYS#1774
We can not have sysmocom-backup-default and sysmocom-backup-data-default
installed at the same time. I don't want to play games with RREPLACE and
RPROVIDES either. So simply use one of the two files. Copy the two more
default values over to the other recipe.
The distro is included after the machine so we can not rely
on the DISTRO_VERSION here. Use the AVAILTUNES and see if the
core2-32 is listed and then pick it.
Create a sysmocom-nitb-rauc-image and create a rauc packagegroup
with the default depdencies and an include file to be used by
other images. This assumes that sysmocom-image.inc has been
included first.
Depending on the upstream version (dora or not) we need to
use either core2 or core2-32. Use a python block expression
to pick this depending on the distro version.
E.g. when using musl and trying to build the smallest possible
rootfs. Existing builds should not start to use these flags as
they would be likely to force a big rebuild.
While copying one build to a new directory we inherit most of
the config but it is better to collect this in a single place
so external contributors build with the same options we use.
With the ro image we have a cheap way to figure out which
files were changed. We can run cmp between the to be backed
up file. This way the stored data will be minimal in the
sense of changed files.
By using initramfs for the rescue we occupy 8mb + ~3mb for the
kernel and can reduce the size of that slot. With glibc it is
difficult to reduce this even further. Let's see if a uclibc
build can help us
In fido the old autoload feature doesn't work anymore and we
need to remove it. Otherwise our sysmobts.service file will
create the /dev/dspl.. /dev/fpgadl... firmware upload files
and we all think that firmware loading is working.
Splitt of the actual rescue root filesystem to
image-rauc-rescue-initramfs and build image-rauc-ubi using
kernel+initramfs (like normal rauc slots).
Signed-off-by: Jan Luebbe <jluebbe@debian.org>
This reboots the system instead of starting a shell on boot problems.
Because the slot will only be marked as sucessfully booted as the last
step in a normal boot, repeated emergency boots will cause a fallback at
the boot chooser level to the other slot.
Testing this is possible by passing 'emergency' via the kernel command
line.
Hopefully, this installs a different /etc/default/udhcpc file on
sysmocom-idu machines. Not sure if adding a machine subdir suffices.
The idea is that the machine-specific subdir symocom-idu should replace
the udhcpc.default file used for the idu packages, only, which has both
resolv.conf mods and default route disabled by default.
The IDU requires that udhcpc does neither touch resolv.conf nor install a
default route. Add a flag to switch off default route installation, the
resolv.conf part is already switch-off-able.
It looks like a classic src != build dir issue of the
utility breaking the build on Yocto-Master. Try to fix
that.
ERROR: Function failed: do_install (log file is located at /home/builder/jenkins/workspace/Yocto-Master/label/OE/build/tmp/work/armv5e-poky-linux-gnueabi/rtl8169-eeprom/v0.0.1+gitAUTOINC+e62e515ce3-r0/temp/log.do_install.2455)
ERROR: Logfile of failure stored in: /home/builder/jenkins/workspace/Yocto-Master/label/OE/build/tmp/work/armv5e-poky-linux-gnueabi/rtl8169-eeprom/v0.0.1+gitAUTOINC+e62e515ce3-r0/temp/log.do_install.2455
Log data follows:
| DEBUG: SITE files ['endian-little', 'bit-32', 'arm-common', 'arm-32', 'common-linux', 'common-glibc', 'arm-linux', 'arm-linux-gnueabi', 'common']
| DEBUG: Executing shell function do_install
| install: cannot stat `/home/builder/jenkins/workspace/Yocto-Master/label/OE/build/tmp/work/armv5e-poky-linux-gnueabi/rtl8169-eeprom/v0.0.1+gitAUTOINC+e62e515ce3-r0/git/rtl8168-eeprom': No such file or directory
| WARNING: /home/builder/jenkins/workspace/Yocto-Master/label/OE/build/tmp/work/armv5e-poky-linux-gnueabi/rtl8169-eeprom/v0.0.1+gitAUTOINC+e62e515ce3-r0/temp/run.do_install.2455:1 exit 1 from
| install -m 0755 /home/builder/jenkins/workspace/Yocto-
From the obn-test config, I wrongly assumed that the inmarsat config should
be static on a vessel. Instead, we need to figure out a way to use dhcp with
linkmon.
The builds of the stable upgrade and the Yocto-Master failed
because of the commit gone missing (no remote is referencing
it). I created a dummy branch to hold this commit.
* We never need the BTS firmware. We never want to do GSM from
within rescue image
* Remove various udev hardware dbs. I think we couldn't care less
about identifying peripheral devices by name and such.
The system should be as small as possible and besides tools to
touch flash and block files, SSHD and rauc there should be nothing
else. We might continue to have opkg on it to allow updating it
The reverse dependency appears broken. Add a comment to
the recipes to warn us when updating. The issue does not
appear to exist in master of Yocto.
Related: SYS#1353
The partition layout needed for rauc is different from the one created
by the existing installer (based on the live image).
An initramfs-framework module for installation to the sysmocom-idu is
added. It is included in a corresponding initramfs image for use on an
USB memory stick.
When using squashfs + overlayfs a root, we need to store the backup
somewhere else and restore it on every boot. Do avoid breaking existing
users, this is implemented in a separate recipe (as suggested by Holger).
We don't use any binary only components and can avoid the performance
penalty.
[hfreyther: The ODU build had changed the tune through
the local.conf already]
Barebox should be built like the kernel. We have to disable the target
userspace tools because they cannot be build using KERNEL_CC (which is
missing --sysroot=).
Move the default file list to a separate file in /etc/sysmocom/backup.d,
which is itself included in the backup. Also create a symlink to the
most recent backup.
* Update libosmocore with GB fixes and endian.h work
* Update libosmo-netif for rtp header changes
* Update OpenGGSN that now has libosmocore logging
* Update OpenBSC with many new features, subscriber queue
rework and more
- remove file with wrong path to ifplugd
recipes-core/busybox/files/busybox-ifplugd.sh
DAEMON=/usr/bin/ifplugd vs. /usr/sbin/ifplugd
- use lambda expression for sysmobts/sysmobsc-kernel to
include busybox-ifplugd or not
- use different defconfig files between master
and dora to enable ifplugd or not
- use different interfaces files between master
and dora to enable auto stanza or not
for the dhcp interface
Fixes: SYS#868 and SYS#1535
[hfreyther: Try to pick an option less likely to break the
Dora support but allowing us to move forward. Not entirely
compile tested due IO load of the system]
The /boot directory is excluded from the squashfs image only, as in that
case it is stored separately. This allows booting normal sysmocom images
as rauc slots.
meta/conf/machine/include/ia32-base.inc:13 (ext3 cpio.gz live)
forces the above image formats that are not relevant to us at
all. Found and done by Jan.
The initramfs-framework from oe-core performs an endless loop of "sleep
3600" when fatal() is called. We need to reboot instead.
Signed-off-by: Jan Luebbe <jluebbe@debian.org>
* Disable ext2/ext3 (as ext4 can be used to mount them).
* Enable AUFS/Squshfs support
* Misc improvements (e.g. ARM optimized AES code, disable
drivers/frameworks unlikely to be used)
Jan prefers to store the fully extended .config file so we will
do it. One advantage is to more easily see what has changed between
two kernel versions.
cp ../build.sysmobts/tmp/work/sysmobts_v2-poky-linux-gnueabi/linux-sysmocom/3.10.80+gitAUTOINC+3122273386-r43/git/.config ./recipes-bsp/linux/files/sysmobts-v2/defconfig
cp ../build.sysmobts/tmp/work/sysmocom_odu-poky-linux-gnueabi/linux-sysmocom/3.2.63+gitAUTOINC+3ed283bea6-r41/git/.config ./recipes-bsp/linux/files/sysmocom-odu/defconfig
cp ../build.sysmobts/tmp/work/sysmocom_bsc-poky-linux/linux-sysmocom/3.10.80+gitAUTOINC+3122273386-r43/git/.config ./recipes-bsp/linux/files/sysmocom-bsc/defconfig
I think the last device we sold was in 2011 and I am not aware
if anyone has used it for the last couple of years. Remove it
to lower the complexity of the builds.
This resolves an error message like "Can't locate bytes.pm in @INC (@INC contains: /etc/perl /usr/lib/perl/site_perl/5.14.3/ /usr/lib/perl/site_perl/5.14.3 /usr/lib/perl/vendor_perl/5.14.3/ /usr/lib/perl/vendor_perl/5.14.3 /usr/lib/perl/5.14.3/ /usr/lib/perl/5.14.3 /usr/local/lib/site_perl /usr/lib/perl/5.14.3 .) at /usr/lib/perl/vendor_perl/5.14.3//JSON/backportPP.pm line 668."
Somehow master and dora generate different checksums for the
same file. Use a different file and start in the "COPYRIGHT AND
LICENSE" and hope it is stable across these two files.
Drop the dependency and more modules. I don't feel like packaging
Types::Serialiser and then their dependencies and so on. The pure
perl based encoding does work and we don't want to run this code
every milli second.
Depend on two more modules to have the following code
emit something:
#use JSON::XS;
use JSON;
my %rec_hash = ('a' => 1, 'b' => 2, 'c' => 3, 'd' => 4, 'e' => 5);
my $json = encode_json \%rec_hash;
print "$json\n";
Besides DBI we need the JSON module. As far as I can see
JSON can use JSON:XS as a fast encoder/decoder. This is
why both libraries need to be provided.
We need DBI and the sqlite3 backend support for a small
monitoring application. Make it available. The files were
taken from meta-perl as of yesterday.
Add the new task to task-sysmocom-feed so it is being built by
default in our CI infrastructure.
make iputils available as well for people wanting to use a real
ping/traceroute/arping.
Related: SYS#1190
Rather than parsing the existing XML or standard output of the nfacct
utility, and then piping that into a perl/python script for formatting
to JSON, we teach nfacct to generate JSON directly. This seems like
the proper way to do it. The patches will be submitted mainline.
[hfreyther: I asked henning to move forward and not test this. We
do not have dhcp/dynamic config in /etc/network/interfaces and the
link will always be up on the idu due it being connected to a switch.
ifplugd is not doing anything useful in this context]
Fixes: SYS#1535
In SYS#659 we wanted that "file" will always log to a persistent
/var/log. After I changed the /etc/default/volatiles/00_core for
the IDU and made journald persistent we still ended up having the
symlink. It appears to be caused by this script.
* When "buffer" is used we don't need to care about /var/log as
it is not being used at all.
* When "file" is set we make sure that /var/log is not a symlink.
In case of the IDU the journald should now be persistent.
Related: SYS#659
The log is spammed with:
May 03 06:24:58 kernel[221]: Loading kernel module for a netw....
May 03 06:24:59 kernel: Loading kernel module for a network ...d.
this is because of an ioctl for an invalid device which leads
to the kernel trying to load a module. This occurs many times
a minute.
I have cherry-picked 1910454fe00cce8b815b1abc0a18a04d3d387ccf which
merged without conflicts. It has not been compile tested.
Related: SYS#1610
For the IDU we currently want to have persistent log files. We
need to fork the volatiles script (installed as 00_core) and it
is already different between dora and current master of poky.
We are missing an sob-idu-qa check that verifies that /var/log
is not a symlink and that /etc/systemd/system/journald.conf is
set to persistent too.
Related: SYS#659
Similar to the T-ODU we would like to keep a persistent log.
This commit is completely untested, rushed and not the final
one (/etc/default/volatiles/00_core most likely needs patching
and we want to limit the size of the log as well).
Related: SYS#659
* add interface for connection to Inmarsat
* remove 'br0' and work directly with 'eth2' instead, to avoid bridges
in vlans of bridges
* bridge the video LAN also into the backbone (for later break-out)
linux-sysmocom_git.bb is used for the BTS, BSC and IDU. Let's
really use 3.10.76 (from 3.10.50 just rebased and not tested
on the device yet). This should fix the Yocto-Master build
tonight.
* the oe udhcpc calls /usr/share/udhcpc/default.script which
calls run-parts -a "$1" /etc/udhcpc.d, so we can install
our script as 60ntp
* using /etc/default/udhcpc-ntp to have the possibility
to disable it on the target, its enabled per default
Related: SYS#1180
9610e792ffde7fb2599d19ca6d2a03230f2043c6 can not be found
in the linux-3.10.76 branch. Just avoid this branch game.
ERROR: Fetcher failure: Unable to find revision 9610e792ffde7fb2599d19ca6d2a03230f2043c6 in branch linux-3.10.76 even from upstream
ERROR: Function failed: Fetcher failure for URL: 'git://git.sysmocom.de/sysmo-bts/linux.git;protocol=git;branch=linux-3.10.76'. Unable to fetch URL from any source.
ERROR: Logfile of failure stored in: /home/builder/jenkins/workspace/Yocto-Master/label/OE/build/tmp/work/sysmobts_v2-poky-linux-gnueabi/linux-sysmocom/3.10.76+gitAUTOINC+9610e792ff-r43/temp/log.do_fetch.16350
NOTE: recipe linux-sysmocom-3.10.76+gitAUTOINC+9610e792ff-r43: task do_fetch: Failed
ERROR: Task 126 (/home/builder/jenkins/workspace/Yocto-Master/label/OE/meta-sysmocom-bsp/recipes-bsp/linux/linux-sysmocom_git.bb, do_fetch) failed with exit code '1
pkg.m4 has been packaged in the main package for a long
while. We certainly don't need this in master anymore.
ERROR: No recipes available for:
/ws/meta-sysmocom-bsp/yocto-master/pkgconfig_0.28.bbappend
It seems like the kernel/package/upgrade/... proceess doesn't deal well
with the fact that some options used to be modules, which are now
built-in. Let's keep it as modules for the time being.
this enables conntrack events, ipset, ebtables, tc
scheduler/classifiers/ematches and much more.
it also disables the ATM code for which I don't think we have any use
case.
* we are the authoritative DHCP server for all local networks
* filter bogus windows DNS queries
* don't respect local resolv.conf (pointing to localhost)
* increase cache size to 1000 entries
Create the /run/network directory for all of our systems. The
proper way with tmpfiles.d ends with a depedency mess that has
no point in being debugged.
My local build hasn't completed yet but we need to rush. Let the
CI test this change and we will see.
Related: SYS#1514
It looks like on my tests with the sysmobts I was saved by
busybox-ifplogd executing ifup again. I couldn't figure out
the dependencies between the LSB networking service and the
tmpfiles.d interaction. So instead of winning a beauty contest
we will need to create the directory by hand.
Cause: SYS#1514
Revert "ifupdown: Package the new file as well"
This reverts commit 15dc869cda.
Revert "ifupdown: Try the workdir for the files we copied"
This reverts commit 1ef790672e.
Revert "ifupdown: Create /run/network through a tmpfiles.d"
This reverts commit e0c7f3c621.
This way the ifupdown application will not only work
for the IDU/ODU but the BTS, BSC and any other machine
we want to build.
(untested OE modification)
- the git version was choosen because it has
some fixes for newer python versions
- test where done with the test file from the github
repository and some unixsocket tests failed
Related: SYS#1082
To be able to talk to the equipment we need to set a route
after the link is up. Assume that no other (e.g. 3G) equipment
is using the same IPv4 address range.
Message-ID: 55279904.9020802@sysmocom.de
As pointed out by Harald the PR is lower than the package
that is already on the device. Bump it to 6 then.
Current:
opkg list_installed | grep ntp
ntp - 4.2.8p1-r5.0
ntp-tickadj - 4.2.8p1-r5.0
ntpdate - 4.2.8p1-r5.0
Import ntp 4.2.8 from meta-networking. Our version has some weird
dependency issues that create build breakages and more.
Remove the pps-tools dependency compared to meta-networking and
leave the ntp.conf as doneby Harald (e.g. drift on /etc and not
in /var...). Now I have not verified the upgrade path but I sadly
merged testing to stable when the ntp changes were comitted on
monday. This means I can't easily test the upgrade.
This new version of ntp has quite a number of changes, e.g.
certain utilities requiring perl, and some programs are now in
/usr/sbin rather than /usr/bin.
The sources are not using autotools and the
build breaks.
ERROR: Function failed: do_install (log file is located at /home/builder/jenkins/workspace/Yocto-Master/label/OE/build/tmp/work/armv5te-poky-linux-gnueabi/i2c-tools/3.1.1-r0/temp/log.do_install.10782)
ERROR: Logfile of failure stored in: /home/builder/jenkins/workspace/Yocto-Master/label/OE/build/tmp/work/armv5te-poky-linux-gnueabi/i2c-tools/3.1.1-r0/temp/log.do_install.10782
Log data follows:
| DEBUG: SITE files ['endian-little', 'bit-32', 'arm-common', 'arm-32', 'common-linux', 'common-glibc', 'arm-linux', 'arm-linux-gnueabi', 'common']
| DEBUG: Executing shell function do_install
| NOTE: make -j 2 DESTDIR=/home/builder/jenkins/workspace/Yocto-Master/label/OE/build/tmp/work/armv5te-poky-linux-gnueabi/i2c-tools/3.1.1-r0/image install
| make: *** No rule to make target `install'. Stop.
| ERROR: oe_runmake failed
| WARNING: /home/builder/jenkins/workspace/Yocto-Master/label/OE/build/tmp/work/armv5te-poky-linux-gnueabi/i2c-tools/3.1.1-r0/temp/run.do_install.10782:1 exit 1 from
| exit 1
| ERROR: Function failed: do_install (log file is located at /home/builder/jenkins/workspace/Yocto-Master/label/OE/build/tmp/work/armv5te-poky-linux-gnueabi/i2c-tools/3.1.1-r0/temp/log.do_install.10782)
NOTE: recipe i2c-tools-3.1.1-r0: task do_install: Failed
In yocto versions before dizzy/master the build was done
in the source directory and when configure was executed
twice in the same directory we would have a build failure
due the symlink already existing. Simply assume that if
the ln fails the link is already there.
Out of disk space situations are handled more early by
OE and missing files will be found by configure.
In dizzy/master OE will clean the build directory before
calling configure so the issue does not apply there.
Fixes: SYS#1170
This is a late fall-out of the priority changes done by
Henning. This recipe is not using autotools and doesn't
support src != build dir builds. So let's not use the
autotools class but simply invoke install ourselves. Let's
lower the default preference so the recipe from yocto master
is picked as well.
ERROR: Function failed: do_install (log file is located at /home/builder/jenkins/workspace/Yocto-Master/label/OE/build/tmp/work/all-poky-linux/ca-certificates/20120623-r0/temp/log.do_install.16890)
ERROR: Logfile of failure stored in: /home/builder/jenkins/workspace/Yocto-Master/label/OE/build/tmp/work/all-poky-linux/ca-certificates/20120623-r0/temp/log.do_install.16890
Log data follows:
| DEBUG: SITE files ['endian-little', 'bit-32', 'common-linux', 'common-glibc', 'allarch-linux', 'common']
| DEBUG: Executing shell function do_install
| NOTE: make -j 2 DESTDIR=/home/builder/jenkins/workspace/Yocto-Master/label/OE/build/tmp/work/all-poky-linux/ca-certificates/20120623-r0/image install
| make: *** No rule to make target `install'. Stop.
| ERROR: oe_runmake failed
| WARNING: /home/builder/jenkins/workspace/Yocto-Master/label/OE/build/tmp/work/all-poky-linux/ca-certificates/20120623-r0/temp/run.do_install.16890:1 exit 1 from
| exit 1
| ERROR: Function failed: do_install (log file is located at /home/builder/jenkins/workspace/Yocto-Master/label/OE/build/tmp/work/all-poky-linux/ca-certificates/20120623-r0/temp/log.do_install.16890)
NOTE: recipe ca-certificates-20120623-r0: task do_install: Failed
ERROR: Task 3639 (/home/builder/jenkins/workspace/Yocto-Master/label/OE/meta-sysmocom-bsp/recipes-extra/ca-certificates/ca-certificates_20120623.bb, do_install) failed with exit code '1'
A lease bound/renewal could lead to the removal of a default
route set by pppd. Apply the same "dev $interface" filter that
would be used when using the "route" command.
Fixes: SYS#1163
The IDU is powerful enough to run the journald. This gives us
some convenienance to have additional output from our services
and we can query them at a later point in time.
This will only be useful for systems that are re-installed
from scratch.
Fixes: SYS#892
UUCP (cu, uux, uu*) attempts to create locks in /run/lock and it
fails. It fails because the current permission of the directory is
0755 but the binary is owned by uucp:root. Modify the legacy.conf
and change the permission to 0775.
I edited the config through make menuconfig and then used
savedefconfig but the diff from our config to a minimal
defconfig is too big. So I enabled the temp sensors and i2c
by hand.
Fixes: SYS#1148
The current 1:0.1+git1+019d2535b6-r9.2 is considered to be older
than the new gpsdate_0.1+git0+2326eee789-r10 file. Change the
PE to benefit from an upgrade.
The ODU has the GPSD for GNSS/GPS listening on a different
port. In hindsight this was a stupid decision but one we can
only fix once we move to generating the config of the system
at boot time.
Fixes: SYS#906
In case the port is not in a forwarding state.. don't just
give up but background and move on. This appears to be needed
for the odu to acquire a lease when STP is used.
For the V-ODU we want to have vlan support as well. This requires
to use the debian ifupdown which means we need a script that creates
/run/network early enough.
The name of the config file was changed and so far
I was hesitant to upgrade the MLO. I think the change
in EEPROM applies to the MLO as well so we should
upgrade it. It has neither been build tested nor tested
on a device.
On eth1 we should also use a network from the 172.16 block rather than
the 10.23 that we used on the sysmoBSC
Also, we now use 172.16.4.0/23 natively on the bridged onboard network,
this is a temporary measure so existing JBs will still connect and get
an IP address before their software is updated to use VLAN6. After they
(and all APs, BTSs, etc.) have migrated, the 172.16.4.0/23 can be
removed.
For systemd and the Debian ifupdown application we need to
have the /run/network directory. The debian init file is
creating it on the fly and we should do the same for now.
networking.service - LSB: Raise network interfaces.
Loaded: loaded (/etc/init.d/networking)
Active: failed (Result: exit-code) since Fri 2015-02-13 17:59:17 UTC; 43s ago
Process: 54 ExecStart=/etc/init.d/networking start (code=exited, status=1/FAILURE)
Feb 13 18:04:47 sysmocom-idu networking[54]: ifup: failed to open lockfile /run
/network/.ifstate.lock: No such file or directory
The creation of /run/network occurs way too late. So we need
to look for another solution.
This reverts commit 86dda7481f.
* DHCP range for 'vlan6' onboard network 172.16.2.0/23
* DHCP range for 'native' onboard network 172.16.4.0/23
* DHCP range for 'admin' port eth1 172.16.6.0/24
* domain name vessel.sob.sysmocom.de
* listen on all interfaces, not just one
* allow admin to specify static leases (ODU) in /etc/dnsmasq.dhcp-hosts
$ ifup br0.6
would go throup _none_up of links.defn and never call the
set_environ for any of the files. Check for environ being
null and then call the set_environ function to initialize
the PATH. This is changing the semantic to allow the iface,
mode and phase to be NULL. I tried to add proper null checks
--exit-on-error is not supported by the busybox run-parts
and the easiest way is to make the ignore-errors option
the default.
I didn't test the change on the device but is looks sound.
* needed to copy over the oe-core kernel-module classes
and modify them because there was no other way to use them and
name the packages kernel-module-backports
Related: SYS#850
This way we can manage bridges and vlans without additional
hacks. This package shouldn't clash with busybox so I need
to install all files with a ${BPN} suffix and then use the
update-alternatives to create the symlinks. The last two parts
were not tested/verified yet.
Ever since the priority of this layer was raised we shadow
some recipes of the default yocto. Now that we don't need
to target edison, remove some of the legacy work.
* the license in README has a new checksum, it now clarifies
to GPLv2
* bump the layer priority otherwise we could not use the latest version,
because the yocto layers have a higher priority
Related: SYS#732
This means that currently only Yocto Dora and Yocto Master are
being targeted. The amount of maintenance to get it to work on
Edison again is too high and it is better to focus on a systemd
only world. Remove the symlinks for yocto-edison.
A quick look and some notes from henning pointed me to wget
which is in the default layers, the sysmobts-2050 which is now
handle dby sysmobts and some older notes that are unlikely
to be addressed.
Okay "make modules" will lead to the kernel.release file
being re-created. So kernel.bbclass of Yocto master needs
to be fixed.
This reverts commit 3cdbc49cbb.
Somehow the do_shared_workdir tries to access a file that
should be created by do_configure but isn't present then.
Add these items to see what happens during the build.
Now reading the command it should probably be add.nwp or
rem.wp but so is life. We can add an alias for add.nwp in a
future version. The boot, recovery and fwup command have been
changed to issue nand rem.nwp before touching the flash.
We want to use backported modules this means we need to make
some thing a module for it to work. Take Daniel's .config file,
run make savedefconfig and update our defconfig.
Related: SYS#850
* The upstream sysvinit file looks good
* We are using systemd service files anyway
* The 2014.63 version doesn't exist and there is no
point in renaming the file.
The sysmobts-mgr is now linking to libgps. This means we need to
DEPEND on gpsd to have this library available. The tracking build
for both dora and master is broken due this.
This addresses the following warning of OpenVPN:
WARNING: No server certificate verification method has been enabled. See http://openvpn.net/howto.html#mitm for more info.
Go back to have our own do_configure. The kernel.bbclass would
execute "oldnoconfig" where we simply executed "oldconfig". Keep
on using our old way (for the legacy of it). E.g. with the question
how we ended up having SMP enabled in the BSC build this might
be the answer.
OE-Core seems to unpack the kernel sources into another
directory now (e.g. to share the sources with builds for
modules). Use the kernel.bbclass kernel_do_configure and
copy the .config to the build directory.
The kernel is now built in a dedicated build directory.
We actually only use the PMIC (TPS65910) RTC in the hardware,
and not the AM335x RTC. Having a driver for a hardware that doesn't
receive actual backup voltage doesn't make sense.
The odu-gpiotool is a small script created by roh to selectively reset
individual peripherals of the system by the administrator. It originates
from ticket https://projects.sysmocom.de/ortelius/issues/655
We use the new sysfs file to obtain the board version number and
configure the port swapping accordingly. If the board_version sysfs
attribute is not present, we default to v2.
We also integrate resetting the hub chip from within this tool right
befeore writing the new register values via I2C.
[hfreyther: We assume that the user might install OpenVPN with
a different priority but that it will always be available in the
runlevel two (e.g. even if the user changed the default runlevel)]
Related: SYS#517
Similiae to the layer code we simplify the machine include to work
for all versions of a 1.5 release. Poky 1.5.4 has been released and
we don't want to continue to play this game.
I have no idea who creates the /run/openvpn directory on Debian.
The path is not in a tmpfiles.d and I don't see the generator
creating it or the service file indicating that it needs to be
created. Place the file with openvpn.NAME.status into the /run
directory which appears to work on the device.
We want to use systemd for managing the lifetime of OpenVPN. Take
the debian generator (which should work with busybox ash) and the
openvpn.service (to inhibit the sysvinit script) and the target
file and install it.
On systems that have ran "update-rc.d openvpn defaults" one need
to manually execute a systemctl enable openvpn.service. This is
not done through a post-inst script and I am not sure if we should
do it. This means there is a danger of ending with a unit that
doesn't start OpenVPN automatically after upgrade!
The scripts/packages have not been tested on a device yet.
We should use a more recent version of the OpenVPN client. Import
it from the meta-networking layer like we did before. The sysvinit
script has just minor whitespace differences and wasn't imported.
We need to test opkg upgrades of OpenVPN through the VPN. We don't
want to kill the tunnel on such upgrade.
One might need to erase the environment but with a default
environment barebox will now send a LLDP during boot up. This
way we can see if a device is on a local link or not.
Fixes: SYS#488
* split out rtl-nic firmware from the big
linux-firmware package
[hfreyther: For whatever reasons the NIC works without
additional firmware as well. I couldn't see in the driver
source when that is the case and which features are missing]
Fixes: SYS#732
[hfreyther: I removed the MACHINE_ESSENTIAL_EXTRA_RDEPENDS as applications
that use the symlinks should rdepend on the package]
[hfreyther: Changed mismatch of tabs/spaces]
We want to enable the watchdog in the bootloader already and
for this to be failsafe we need to have a kernel that can boot
and get ethernet without any additional modules. Now systemd
would not be able to trigger the watchdog and the system would
reboot within 60 seconds. I am not sure how the kernel upgrade
will go and we need to try this.
Related: SYS#374
We would leak all the results for all the scans. At some point in
time the kernel OOM would kick-in and kill the process. Enable
building with -ggdb3 for debug symbols and remmeber the old list
head so we can free it.
==32240== 756 bytes in 3 blocks are definitely lost in loss record 7 of 8
==32240== at 0x40291CC: malloc (in /usr/lib/valgrind/vgpreload_memcheck-x86-linux.so)
==32240== by 0x804BCF5: iw_process_scanning_token (iwlib.c:3059)
==32240== by 0x804BCF5: iw_process_scan (iwlib.c:3255)
==32240== by 0x804BEAF: iw_scan (iwlib.c:3310)
==32240== by 0x8048FD8: scan_wifi (wifi2udp.c:72)
==32240== by 0x8048DB2: main (wifi2udp.c:143)
The notion of "task" has been deprecated for a long time because
the word is used for different things. The new word is "packagegroup"
which describes the usage of "task-*" in a more obvious way. Dora
and master builds can fully use packagegroup-* for everything and
for edison we would need to spend some time to add a provides for
packagegroup-* to the task- group. I don't intend to work on the
edison part right now.
The notion of task- has been deprecated a long time ago. Call
it packagegroup now and for "packagegroup-osmocom" provide the
compability. This has not been compile tested.
The edison build will require a bbappend to provide packagegroup
now but I will do this once I need to build for Edison again.
We need a jffs2/ubi depending on the age of the hardware and
maybe a tar.bz2 so we can easily inspect an image without
playing with mtdram. We almost never need a ubifs and certainly
not a cpio.
The DISTRO_FEATURES_INITMAN variable has been removed, use
VIRTUAL-RUNTIME_init_manager which is still present. We have
already defined the later in our poky builds
systemd-compat-units does not install any files anymore and
the machineid and other scripts are gone. We want to continue
to install the alignment file though.
This has not been tested on dora systemd yet.
In dora there is no autotools-brokensep bbclass. So let's just
set the build directory to the source directory ourselves. I have
not built the package for dora though/
Poky dizzy is more strict with -I/usr/include and the old
tcpdump could pick-up "/usr/bin/pcap-config" from the host
and fail. Just upgrade to a newer version of tcpdump and be
done with it.
* the entry "option subnet" takes the
subnet mask not the subnet address,
while the busybox dhcp-client was
fine with it, the isc-dhcp-client complained with
Error: an inet prefix is expected rather than "10.23.24.113/10.23.24.0".
WARNING: QA Issue: ortp rdepends on libcrypto, but it isn't a build dependency? [build-deps]
WARNING: QA Issue: ortp rdepends on libssl, but it isn't a build dependency? [build-deps]
We used asterisk to be able to speak SIP with the outside. LCR
has gained this capability in the past and we haven't used asterisk
in more than two years.
Asterisk doesn't seem to compile with master/dizzy and there is
little point in figuring out why.
* bump the PRINC
[holger: Depending on some global settings /var/log will be a
symlink to "volatile". But when we log to a file we want it to
be persistent. So make sure /var/log is not pointing to volatile
but a real directory]
The concept of directdisk image is overhauled as we are now
having a initramfs that can flash the next image through USB.
The dedicated E1/IP images are rarely to never used and did
probably bit-rot. For the next time we provide such a system
we can simply (write a script to) install the additional feed.
CONFIG_IFUPDOWN_UDHCPC_CMD_OPTIONS was set to '-n' which means
"Exit with failure if lease cannot be immediately negotiated."
What we'd rather want is for udhcpc to continue to run until a lease can
finally obtain. Otherwise we have a race condition in case the DHCP
server is not reachable for a few seconds during system boot, and end up
having a system that is permanently without an IP address.
This fixes ORT#639
The line doesn't have the signalling type set yet. It will be set
_after_ the rsl connection is considered up. This is addressing
RSL Connection failures like:
<0001> oml.c:1026 Rx IPA RSL CONNECT IP=1.2.3.4 PORT=3003 STREAM=0x00
<000f> input/ipaccess.c:921 cannot create RSL BTS link: trx_nr (0) does not refer to a signalling link
<0001> oml.c:1030 Error in abis_open(RSL): -22
<0001> oml.c:341 Sending FOM NACK with cause Message cannot be performed.
Without inheriting pkgconfig the PKG_CHECK_MODULES macro can not
be expanded in dizzy. Fix it. While being there use "autotools"
for the pcu and not autotools_stage.
We have changed the strategy to support multiple BTS with a single
build but don't remove support from files that are used with Edison.
Linux 3.10 and systemd were not used with edison though. So we can
remove it. As the sysmobts-2050 extended the sysmobts-v2 build most
of the -sysmobts-2050 changes are not needed anyway.
Fix the build by using a newer version and change the PR to be
a proper version number. I havent' checked if upgrading is possible
but is not a big concern right now.
This variable doesn't exist on dora and will not break things.
Fixes:
ERROR: Please ensure module rtfifo is listed in KERNEL_MODULE_PROBECONF since module_conf_rtfifo is set
We are currently using 1.5 (dora). Copy the bbappends to the
yocto-dizzy directory and update the version numbers. For busybox
enable all features that were enabled between dora and origin/master.
For opkg_svn.bbapend. In dizzy there is a version attached to it.
This means we need to remove it from the yocto-shared directory
and keep it everywhere else.
WARNING: FILESEXTRAPATHS-variable, must always use _prepend (or _append)
type of assignment, and don't forget the colon.
Please assign it with the format of:
FILESEXTRAPATHS_append := ":${THISDIR}/Your_Files_Path" or
FILESEXTRAPATHS_prepend := "${THISDIR}/Your_Files_Path:"
in your bbappend file
Creating a new file in /etc/sysmocom/backup.d/ that ends with .backup
will be read during the backup and then be restored. This way we can
easily extend the files to backup.
This is using UDP similiar to gps2udp. Holger has pushed for a
simple solution like this to be replaced by a proper application
that will use AMQP.
Fixes: ORT#586
* take the baudrate for the grub serial console setting
from the kernel-cmdline which the system was booted from
* better parsing of the options which the install-script
gets from init
* Video Mode, VGA mode are not used right now and make it
empty right now.
* when restoring links e.g. for systemd files
the links will be created with content// prefix
* use transform instead of strip to remove
the "content/" prefix
* bump PR
Add post install script that will mangle the config to be the
right one for the given machine. This will be executed once
after the post installation. In case of the getty change the
system must re-start once to be fully usable. E.g. for the
sysmoBTS2050 the PA might not be powered up on first boot.
The sysmobts-2050 machine does inherit from the generic bts so
we do not need an exact copy of the configuration file.
The watchdog is not used in dora or later images as this functionality
is integrated into the systemd.
Related: SYS#559
We want to have a single image that will do the right thing on
the various BTS models we have. This means we will install the
sbts2050 utilitiy (it clashes with sysmobts-mgr anyway) and the
gpsd with a default config for it.
The sysmobts-v2 config file was for a specific usecase and is
obsolete for all intents and purposes. Make it the default
Related: SYS#559
The control interface has been moved to libosmocore and a new
library has been created for it. The ipaccess code has moved
from libosmo-abis to libosmocore and new defines for the VTY
and CTRL ports are now available inside libosmocore
We need to replicate the other services that were included with
*.service. Install an additional alignment.service that is required
for ARM. The actual service file is from Daniel and copied from
the postinst and sysmobts service.
I wish I could have made it more simple but at least only compare
the 1.X of the poky version as minor versions generally do not
upgrade the core components
Having:
osmo-bts requires sysmobts-firmware > 3.11
sysmobts-firmware requires osmo-bts > 0.4.2
is something opkg can't resolve in one go. One needs two calls to
the opkg command to end up with a fully upgraded version. Make sure
that we require/pull newer firmware when installing a new version
of osmo-bts.
* Add sysmobts-firmware 3.11. The firmware will work with the
previous osmo-bts software and this one (as we don't use the
new primitives)
* Build newer version of osmo-bts with support for online
calibration and CMR fixes.
Enabling the "systemd" feature will enable the socket activation
feature used by systemd. This way we can (re-)start the GPSD based
on the socket activation.
If the year is smaller than < 2014 an initial date will be set. This
has been manually verified by copying ntpdate to ntpdate.off and making
sure that the VPN is coming up. Due the short SSL lifetime of CACert
we will need to update the script once in a while. But our VPN should
now always come up.
The TI Davinci DM644x and the TI Sierra AM335x both feature a
watchdog. Enable it on the AM335x as well. The next part is that
we can enable the watchdog for the shutdown. E.g. in case the
system does not reboot after the shutdown.
The shutdown is executed last and only after the shutdown target
has been reached. For time timeout we need to use the same interval
as the watchdog (as both drivers don't allow to change the timeout
interval).
Henning got really confused that the "a.b.c:1234" he passed to
gps2udp showed up as "a.b.c 1234" in the kernel cmdline. Use
strdup to take a copy of it.
Fixes: SYS#424
systemd doesn't look in /etc/rcS.d for init scripts so we never
start networking unless we are using ifplugd. We need to check
how this behves on the sysmobts and sysmobsc in case the network
is being started twice.
Build a version of barebox that has support for NAND flash and
ubifs. Load the kernel from the device. One needs to flash the
*mlo*.img into the xloader partitions and the other image into
the barebox partition.
erase /dev/nand0.xload.bb
erase /dev/nand0.xload_backup1.bb
erase /dev/nand0.xload_backup2.bb
erase /dev/nand0.xload_backup3.bb
cp /mnt/mmc0.0/MLO /dev/nand0.xload.bb
cp /mnt/mmc0.0/MLO /dev/nand0.xload_backup1.bb
cp /mnt/mmc0.0/MLO /dev/nand0.xload_backup2.bb
cp /mnt/mmc0.0/MLO /dev/nand0.xload_backup3.bb
This doesn't work like this. We need to know the _exact_ PV/PR
of the package and we don't. Specially with the PR server in dora
and later we can not know the version.
This reverts commit 767d7b6939.
So far we have used chrpath to fix the rpath we don't want but
this creates issues with supporting both dora and edison (we need
to depend on different recipes). Instead just force the RPATH to
be off. This can be done by setting the internal _RPATH variable.
I found this in Tool/gnulink.py.
This upgrades the firmware from 3.6 to 3.10. This is a significant
firmware upgrade and it changed ABIs too. Try hard to fix the
dependencies as much as possible. This requires the DSP headers
files to be of version 3.10 as well but we can not have versioned
DEPENDS.
The code has not been finished and doesn't meet the quality
standards we have for a product. Comment the sysmobts-mgr
related files that will be back soon.
osmo-sgsn was installing a sysvinit script but it wasn't inside
the INITSCRIPT_PACKAGES so it would not be started by default.
Use the already existing INITSCRIPT_PARAMS/INITSCRIPT_NAME variables
for the osmo-sgsn.
Fixes: SYS#366
We need to reduce the usage of ttyS0 to not conflict with the
sysmobts-mgr that opens and controls the power amplifier. This
is done by instructing systemd by config file to not log much
and disable the getty target.
* archiver discard the source providing
for all native and shared-work recipes,
so we whitelist gcc-cross, which covers
the source code for the runtime libgcc
package
Fixes: SYS#372
* to be able to mount vfat formatted
usbsticks we need the nls subsystem
* providing nls as modules is a bad idea because
you need to add them at image creation
* better include them into the kernel,
like the kernel for bts and bts-2050 does
Let's flash multiple copies of UBL and U-Boot to cope better with
NAND bits flipping in these sectors. To fit multiple copies of
U-Boot into the flash we had to move the rootfs a bit to the end.
* setting APPEND in the image file overwrites
the APPEND from boot-directdisk.bbclass in edison, which
creates the syslinux.cfg without "root=" stanza
* for dora we do not use syslinux or the image-directdisk
so revert this paticular changes
Fixes: SYS#373
Daniel noticed that on upgrade his /etc/osmocom/osmo-bts.cfg
was overwritten. This was due the addition of the sysmobts-mgr
config file and using an assignment instead of an append.
We will change the MTD partitioning for our devices soon. The
kernel needs to honor the cmdline given by the kernel for that.
The rootfs will move a few pages to the back with newer bootloaders.
If we don't provide a DNS server via DHCP to the BTSs, then their
ntpdate will not succeed in contacting pool.ntp.org (resolver failure),
which in turn means they have the wrong date, which again in turn leads
to failing opkg update or openvpn certificate verification failures.
Update to current master of lcr, which includes support for AMR
and RTP-bridge. Master openbsc doesn't support RTP-bridge yet,
but it will work with a transcoding LCR getting all TCH frames
in all codecs (HR/FR/EFR/AMR) via MNCC.
The .pc files were already covered by the normal -dev package
glob. No need to do more work here.
Addresses:
WARNING: Variable key FILES_${PN}-dev (${includedir} ${FILES_SOLIBSDEV} ${libdir}/*.la ${libdir}/*.o ${libdir}/pkgconfig ${datadir}/pkgconfig ${datadir}/aclocal ${base_libdir}/*.o ${libdir}/${BPN}/*.la ${base_libdir}/*.la) replaces original key FILES_gpsd-dev ( ${libdir}/pkgconfdir/libgpsd.pc ${libdir}/pkgconfdir/libgps.pc).
Currently OpenBSC and LCR are not compatible with each other.
So far we have only included patches that add new defines and
do not change the wire format of messages. Just revert it so
we can have LCR talk to OpenBSC
While working on the OpenBSC/NITB schema migration I experienced
crashes and traced it down to libdbi-drivers. It was possible that
a special string quote each character and then adding '"\0' would
result in out of bounds write.
These provide us with a nice way to provision/install the BSCs
with grub. This way we can easily upgrade the kernel on the BSC
as well. It is a fix for SYS#75.
We do not intend to use u-boot right now but at least this
version will build. In the first version we will use the
barebox images provided to us by the producer of the SoM.
This is fixing max_power_reduction for revC hardware and applies
a new max_power_reduction when it arrives through OML and the
system is already initialized.
The name/structure of the NAT configuration was changed. Catch up
and put the new names into the old defconfig too. It is an ugly hack
to have it work with v3.2 and v3.10 kernels. The sysmobsc has been
fixed independently.
This might require to rebuild the freshly built openggsn to
fix the libgtp0 depedency. I had this issue on Dora without
a PRServer. I am not sure it applies to edison or other configs.
Bump the PR of OpenBSC to force a rebuild of the SGSN.
Fixes: SYS#176
We had an upgrade issue where using newer firmware with older
osmo-bts binaries didn't work. Make sure that the shiny new 0.4.0
binaries are used with firmware starting from 3.2.
Addresses: #158
When disabling fb, directfb, x11, widgets and OpenGL qtbase will
still complain that no QPA backend is available. Always add the
-no-qpa-platform-guard parameter
The license agreement doesn't mention Nokia anymore after Qt was
acquired by Digia.
Signed-off-by: Jonathan Liu <net147@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* use seperate defconfig, because of option changed from 2.6.39 to 3.10
* use python magic to switch between linux and linux-sysmocom
* add sysmocom-bsc to compatible machines for linux-sysmocom
* add new confiugration parameters for syslinux
* drop 0003-Fix-wrong-calculation-of-viewPort-for-transitions.patch
because it's already applied in upstream commit:
commit d765ad8e3dabe44f276db9084b1d16164c4535e7
Author: Florian Hänel <florian.haenel@basyskom.com>
Date: Wed Jun 5 19:23:32 2013 +0200
Subject: Fix wrong calculation of viewPort for transitions
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* in some components there was just
Bump MODULE_VERSION to 5.2.2
commit but bump them all for consistencty
* we're testing newer revisions in order to narrow some QML issues
we're seeing in runtime
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* there is no kms around
* mesa is the default implementation for (lib)gbm (see config.tests/qpa/kms)
* in configure we find:
| if [ "$CFG_KMS" = "yes" ]; then
| if [ "$CFG_EGL" = "no" ]; then
| echo "The KMS plugin requires EGL support and cannot be built"
| exit 101
| fi
| fi
Signed-off-by: Andreas Müller <schnitzeltony@googlemail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Use gstreamer for 1.x, gstreamer010 for 0.10.x.
Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
The dependency on qtsvg seems to be a leftover from an old version.
Version 5.2.1 of qtdeclarative doesn't include any reference to it.
Depend on qtbase in case qtxmlpatterns gets disabled.
Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Pango doesn't seem to be a drop-in replacement for ICU, so
using it as a dependency for the -icu option won't work.
QtWebKit has a hard dependency on icu.
Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
This adds a new image feature for easier remote debugging with
QtCreator.
The qtcreator-debug needs to be added as a valid item for image
features and having it inside the populate_sdk_qt5 is too late. User
will see a parser error as OE-Core images does not inherit the classs.
The possible workaround for it is to include it inside layer.conf file
until a kind of "bbclassappend" support is in place.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Tested-by: Giulian Vivan <giulian@ossystems.com.br>
This allow for easy installation of needed packages for remote
debugging using QtCreator. The qtdeclarative-plugins package provides
the needed infrastructure to remotely debug QML application.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Tested-by: Giulian Vivan <giulian@ossystems.com.br>
Since commit 8da5017712, libQt5*.so files
are included in the nativesdk-qtbase-tools package. However, this together
with the fact that the binaries are placed in a subdir "qt5" of bin will
confuse the Debian-style package renaming logic in debian.bbclass, causing the
resulting ipk package to be created as "nativesdk-libqt5core5".
Signed-off-by: Jacob Kroon <jacob.kroon@mikrodidakt.se>
Acked-by: Otavio Salvador <otavio@ossystems.com.br>
This currently requires the meta-ti layer but we will make sure
that nothing from it will be required in the end. The kernel is
based on linux-3.2 as this is the "default" TI kernel that supports
the SGX (not that we care)
The QtCreator is quite broken regarding finding the mkspecs files; so
to workaround it we must to add a symlink to it. The workaround was in
place but in the wrong sysroot, it must to be in the /native/ one.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Acked-by: Jonathan Liu <net147@gmail.com>
The qmake class now respects the QT_DIR_NAME setting when setting the
headers path, so this change is not needed anymore.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
The qmake class now respects the QT_DIR_NAME setting when setting the
headers path, so this change is not needed anymore.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
When looking for Qt5 headers we ought to use the Qt5 specific path. In
case the distribution wants to have a flat tree it can set QT_DIR_NAME
and it will still works as expected.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
* egl was building fine, but glx was failing to find
qwaylandintegration.h
* one example cannot be built without gles2, so make it
conditional
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* at least in 5.2.1 tests aren't enabled by default:
QT_ALL_BUILD_PARTS=" libs tools examples tests "
QT_DEFAULT_BUILD_PARTS="libs tools examples"
so without -make option they weren't enabled even with tests in
PACKAGECONFIG
* add options for other build parts just for completeness
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
If qtbase is configured with openssl support then the
qtwebkit browser example apps require CA certificates.
Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Configure qtbase with -openssl-linked (instead of -openssl) to ensure
that run-time dependencies on libcryto and libssl are detected.
Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Also add -developer-build to PACKAGECONFIG
Signed-off-by: Mikko Levonmaa <mikko.levonmaa@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* This way we don't mangle ${B} and can repeat do_install if necessary. And
previously it would require running do_compile again to restore the proper
state of ${B}.
* Need to bump PR, since do_compile isn't changing, but needs to re-run to
re-populate qmake-real binary.
Signed-off-by: Denys Dmytriyenko <denys@ti.com>
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
* Jonathan Liu reported that qtbase-native was failing to build for him
and it looks like qtbase-native was trying to link against icu from his
host (instead of icu-native because it isn't in DEPENDS)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Forcing ARM_INSTRUCTION_SET to arm when building qtbase no longer
seems to be required.
Confirmed by forcing ARM_INSTRUCTION_SET to thumb and building
qtbase 5.1.1 and 5.2.1 tuned for both armv4t and cortexa9thf-neon.
Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
I'm trying to get the .la-files for Qt5 dev-packages correct.
What I've come up with so far is attached in the patch (to be applied in
top of "jansa-qt5-5.2.0", b8a236df79)
With the patch applied I can still build both an image and an SDK, and
also the resulting SDK can be used to build my QT5 autotools projects. But
I get these warnings when building for instance "qtxmlpatterns":
WARNING: QA Issue: qtxmlpatterns: The compile log indicates that host
include and/or library paths were used.
Signed-off-by: Jacob Kroon <jacob.kroon@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* qtwayland: Use default LICENSE and LIC_FILES_CHKSUM
* license texts were added in qtwayland commit 26bdc66fe6c4499332b2ead886a806d09aa07a8a
* qtdeclarative, qtserialport: Update QT_MODULE_BRANCH
* SHA-1 of v5.2.1 tag is only in release branch
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* this is useful for using newer meta-qt5 with older oe-core
release
Signed-off-by: Gabe Johnson <gjohnson@agleader.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Fixes:
ERROR: QA Issue: nativesdk-qtbase: Files/directories were installed but not shipped
/opt/oel/13.0+snapshot/sysroots/i686-oelsdk-linux/usr/share
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
We must to fix the installation paths as did in nativesdk-qtbase.inc
and remove the files unused from the packages. This fixes the
packaging errors and allows for a successful build for SDK.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Acked-by: Denys Dmytriyenko <denys@ti.com>
When building nativesdk binaries we ought to rely in the native
mkspecs.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Acked-by: Denys Dmytriyenko <denys@ti.com>
The 7d8f8d6 change (nativesdk-qtbase.inc: Remove unused contents)
introduced a build failure that was not catch in my test. This patch
fixes it avoiding removing the headers and mkspecs files which are
need for the nativesdk-qttools recipe.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
In case changes are done in do_install and the sstate checksums
change, the task would fail as the qmake binary has already been
replaced. Avoid this error checking for the original filename.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
This allow the toolchain generation using '-c populate_sdk' as it
includes the need host libraries into the SDK.
Signed-off-by: Denys Dmytriyenko <denys@ti.com>
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Allow easy inclusion of target packages for the Qt5 SDK or external
toolchain.
The list of the target package is based on the Arago Project, which
has been contribute by Denys Dmytriyenko <denys@ti.com>.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
This is the latest stable kernel. I had hoped some crashes to
go away with it but it turns out to be a compiler issue that
we backported a patch for separately.
We don't want to launch the dbus-daemon and we did this by modifying
the RDEPENDS_${PN}. Back then this was okay as dbus was the only
dependency but now kmod got added to the recipe as well. Use the new
language construct to just remove a specific value from the rdepends.
Paul Eggleton was nice enough to point that out.
Fixes: SYS#231
Make sure that there is at least a basic depmod in the image. This
is needed as fallback for kernel upgrades that execute depmod in the
postinst script. The dora busybox will always have it so we should
have it too. Sync the defconfig with dora.
Addresses: SYS#231
This patch caused a linker error:
| linking ../../bin/jsc
| [...]ld: cannot find -licuuc -licui18n
The linker command-line contains:
| [...] -licuuc\ -licui18n [...] -licui18n -licuuc -licudata [...]
So the problem gets caused by the backslash, making the actual library
name "libicuuc -licui18n.so". This may have been caused be the double
quotes used in the patch. However, since both libs are already part
of the command-line, the patch can just get dropped. The patch indicates,
that it was created for a beta version, so I assume that this was
fixed before the 5.2 release.
Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Fixes the same build error as 671313d, but for the -native recipe.
Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* it doesn't support gst-1.x like qtwebkit yet, disable it by default to
build images without gst-0.10 where possible.
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
this does not build much yet but I was looking for qdbus
Signed-off-by: Andreas Müller <schnitzeltony@googlemail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
libosmo-sccp does not install any dynamically shared libraries
so let us install/depend on the staticdev package installing the
.a files and the -dev package.
* without syncqt cmake config files have empty PRIVATE_INCLUDE_DIRS
remove pregenerated header files and create .git dir to let qmake
call syncqt and generate headers like when building from git recipes
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Nothing in qtbase uses gstreamer so the configure option was removed on the way
to 5.2 [1]. Fixes:
| -no-gstreamer: invalid command-line switch
[1] e25a4faf4a
Signed-off-by: Andreas Müller <schnitzeltony@googlemail.com>
* drop 0027-Fix-misaligned-selection-region-with-text-when-cente.patch
resolved in upstream commit 5d8a882c11201a29475c5ea71cfb76c9de6573f5
* drop 0020-Use-BGRA-extension-in-bindTexture.patch
resolved in upstream commit e1325cf26e146b68725cc1a0a02b274ce3dfbe5c
* drop 0008-wayland-scanner-disable-silent-rules.patch
resolved upstream in:
commit 2ff2a7c32d76b9e58b800f12469f112cfdb6ad3c
Author: Jan Arne Petersen <jan.petersen@kdab.com>
Date: Fri Jul 19 14:35:19 2013 +0200
Fix wayland-scanner to work with CONFIG+=silent
* squash to match more with structure of
https://github.com/meta-qt5/qtbase/tree/stable
* qtmodules: bump SRCREVs for 5.2.0 tags
now all modules using qt5-git should be newer than any 5.1.* version
* qtbase: fix paths in packageconfig *.pc files
include- and lib-paths contained build-sysroot paths
* qtwebkit: Explicitly add ICU libraries to LIBS
fails to build without this
* qtjsbackend: remove for git versions
Found in [1]:
Qt Qml is now using its own built-in Javascript engine and does not depend on
V8 anymore. As such the QtJSBackend shared library has disappeared.
[1] http://blog.qt.digia.com/blog/2013/09/30/qt-5-2-alpha-available/
* qt5: Upgrade 5.1.1 recipes to 5.2.0 to match git recipes
qtjsbackend is now completely gone
it allows to share more .patch files and configuration in .inc again
Signed-off-by: Andreas Müller <schnitzeltony@googlemail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* since this commit:
commit e0debe138cb59c140ebbc69755d36028d2b89459
Author: Allan Sandfeld Jensen <allan.jensen@digia.com>
Date: Thu Feb 28 13:42:27 2013 +0100
Subject: Switch QtWebKit to GStreamer 1.0.
qtwebkit supports 1.0 and will use that by default when found
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
If Qt5 (qtbase) is machine specific, then all packages depending
on qtbase will also be machine specific, because the (initial)
qtbase configuration becomes part of Qt5/qmake, thus affecting
builds of packages built against this specific version of qtbase.
Since it is not feasible to add PACKAGE_ARCH=... to each and every
recipe that is built using qmake and depends on Qt5, we introduce
a new variable, QT_PACKAGES_ARCH, that can be set by the machine
configuration as needed.
We have to be careful though to only change the architecture of the
resulting packages if we're not compiling a 'native' package.
Signed-off-by: André Draszik <andre.draszik@linaro.org>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Previously license files were missing for qttools module, so we used to point
to generic files with same license.
Signed-off-by: Denys Dmytriyenko <denys@ti.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* systemd has no way to set flexible enviroment variables,
though setting or dropping variables based on other settings is not
possible
* EnvironmentFile option in systemd can only have name value pairs no logic
* Best solution for this problem found at
http://www.mail-archive.com/systemd-devel@lists.freedesktop.org/msg05091.html
using a shell before starting the service
* move the parameter logic from sysvinit startup file to syslog-startup.conf
and source syslog-startup.conf from the busybox-syslog default file
* /etc/syslog-startup.conf is now the only config file
* works with bash and busybox-shell
* bump PRINC
* The Linux 3.2.X kernel has some funny issues with GCC 4.8 in the
fs/namei.c look up code. It is easier to move to 3.10 than to fully
debug the issue.
* disable the mISDN loop patch for v3.10
* use the hack to make v3.2 compilable for edison and v3.10 for dora
Use the native systemd service file to avoid running a shell script
that is ran from within the screen. This code expects the pcu to have
a .service file itself.
Provide custom systemd config files to enable the watchdog handling. For
systemd we depend on the image containing and using systemd as init script.
In case we build images without systemd we will need to install the
old watchdog. Maybe it is time to create a MACHINE_FEATURE for watchdog.
Do not attempt to start/stop the application as part of the upgrade.
This might conflict with systemd .service files that override the
LSB script. The path of systemd has changed so the path was updated
in the script. A better fix would check if the systemd cgroup is mounted
as this is a better indicator for systemd.
systemd can cope without having a dbus-daemon present. This is the
way we want it to operate to conserve a bit of memory. Do not install
the dbus daemon by default. Not having a dbus daemon is breaking the
systemd color output right now but this is something that should be fixable.
Master has moved forward and we need to move the bbappends into the
yocto-edison folder. For the task-core-boot it has been renamed to
the packagegroup-core-boot and we do not need to increase the version
number for master.
Some examples require qtwebkit-qmlplugins (flickerview, youtubeview),
and this dependency is not automatically added. Without this patch,
and without manually adding qtwekbit-qmlplugins, we get errors like
this:
qrc:///flickrview.qml:42:1: module "QtWebKit" is not installed
Signed-off-by: Nicolas Dechesne <nicolas.dechesne@linaro.org>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
When we're building maliit with it's input context plugin enable it gets installed into
${libdir}/qt5/plugins/platforminputcontexts which couldn't be added to ${PN} cause it's
already in ${PN}-dev and the only part which should go into ${PN}-dev is
${libdir}/qt5/mksepcs and nothing else from ${libdir}/qt5.
Signed-off-by: Simon Busch <morphis@gravedo.de>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
The values from maliit-defines.prf are only used at runtime and not at compilation time so
we don't have to replace them with the proper staging paths.
Signed-off-by: Simon Busch <morphis@gravedo.de>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
this might help us in some installations where there are multiple
back-hauls and we want to base routing decisions not only based
on the destination IP.
* system freetype works only together with fontconfig
* building qbasicfontdatabase fails with system freetype and without
fontconfig or with freetype completely disabled
| /home/jenkins/webos-ports/workspace/webos-ports/tmp-eglibc/sysroots/a500/usr/include/ft2build.h:56:38:
fatal error: freetype/config/ftheader.h: No such file or directory
| #include <freetype/config/ftheader.h>
| ^
| compilation terminated.
| make[2]: *** [.obj/release-shared/qbasicfontdatabase.o] Error 1
| make[2]: *** Waiting for unfinished jobs....
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* it was explicitly disabled in
commit 1b14686657
Author: Philip Craig <phil@blackmoth.com.au>
Date: Wed Jun 26 02:13:02 2013 +0000
Subject: qtbase: disable fontconfig support by default
* and then accidentaly enabled with PACKAGECONFIG changes
commit ebdd83f6b7
Author: Martin Jansa <Martin.Jansa@gmail.com>
Date: Sat Jul 20 00:02:57 2013 +0200
Subject: qtbase: explicitly enable or disable all deps
* Thanks to Nicolas Dechesne for reporting it
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* commit 105ad2a18e
Author: Eric B<C3><A9>nard <eric@eukrea.com>
Date: Mon Oct 7 22:54:12 2013 +0200
Subject: qtbase: update to 5.1.1
removed it from SRC_URI in qtbase.inc, but it's still needed for
qtbase_git
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
In the sysmoBTS 2050, we want to configure the 1PPS output of the
trimble to be active only if it has at least 1 satellite in view,
and want to make the receiver output its time information in UTC,
rather than GPS time.
Note: binaries will go into ${PN}-examples, which is qtwebkit-examples-examples
Signed-off-by: Denys Dmytriyenko <denys@ti.com>
Signed-off-by: Eric Bénard <eric@eukrea.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
jsondb is not listed a a dependencies on qtlocation in sync.profile
and was removed in more recent versions as qtjsondb is no more
maintained
Signed-off-by: Eric Bénard <eric@eukrea.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
- patch 10 was update
- patch 19 & 25 are merged so no more needed
Signed-off-by: Eric Bénard <eric@eukrea.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* Imported from rev 13ae5105ee30410136beeae66ec41ee4a8a2e2b0
This is the first step to move these recipes over to qt5.
* maliit-framework-qt5: Build with qt5
Switch to new github repo
Update SRCREV and remove patches which now longer apply
Disable doc generation and xcb
Fix packaging
* maliit-plugins-qt5: Build with qt5.
Switch to new github repo
Update SRCREV and PV
Fix packaging
* maliit-*-qt5: Adjust MKSPEC path
* meta-framework-qt5: Cleanup packaging and enable dbus activation.
Move pkgconfig files, headers and spec files into dev.
Cleanup packaging rules
Enable dbus activation to start maliit server on demand
* maliit-plugins-qt5: Set MALIIT_PLUGINS_DATA_DIR to avoid path leakage from host
If not overriden here MALIIT_PLUGINS_DATA_DIR would point to some host
paths resulting in e.g. qml files not being found.
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
sbts2050-util is a small binary-only userspace program used to
interface with the sysmoBTS 2050 board management controller,
i.e. to enable/disable PA power, read power consumption/temperature
and the like.
This tool is going to disappear once the respective features have been
included in the (open source) sysmobts-util / sysmobts-mgr. Please be
patient meanwhile.
The build differs in that it doesn't use a getty on ttyS0. In the fuutre,
a modified kernel serial driver as well as a 2050-specific utility
for managing the supervisor processor will follow.
If you perform a "bitbake -c devshell <some qt5 recipe>" and then try "which
qmake" in the devshell, you'll find your host's qmake is found (which is most
likely to be based on qt4) rather than the one that is built as part of
meta-qt5. The reason for this is that the qt5 items are installed to
${STAGING_DIR_NATIVE}/usr/bin/qt5 and not ${STAGING_DIR_NATIVE}/usr/bin. The
reason for the extra "qt5" directory at the end is so that qt4-based items
that are built in OE can live side-by-side with things from meta-qt5 without
interfering with each other.
This patch prepends ${STAGING_DIR_NATIVE}/usr/bin/qt5 to the devshell's PATH
so that the qt5-based native tools which are built as part of meta-qt5 will be
found before the host's.
Signed-off-by: Trevor Woerner <trevor.woerner@linaro.org>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Building qtwebkit causes a QA issue such that QtWebPluginProcess and
QtWebProcess contain bad RPATHs which point into the build location. This fix
adds a patch to not include the rpath.prf which causes this problem.
Signed-off-by: Trevor Woerner <trevor.woerner@linaro.org>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
| CMake Error at tmp-eglibc/sysroots/qemuarm/usr/lib/cmake/Qt5Core/Qt5CoreConfig.cmake:27 (message):
| The imported target "Qt5::Core" references the file
|
| "/qmake"
|
| but this file does not exist. Possible reasons include:
|
| * The file was deleted, renamed, or moved to another location.
|
| * An install or uninstall procedure did not complete successfully.
|
| * The installation package was faulty and contained
|
| "tmp-eglibc/sysroots/qemuarm/usr/lib/cmake/Qt5Core/Qt5CoreConfigExtras.cmake"
|
| but not all the files it references.
|
| Call Stack (most recent call first):
| tmp-eglibc/sysroots/qemuarm/usr/lib/cmake/Qt5Core/Qt5CoreConfigExtras.cmake:6 (_qt5_Core_check_file_exists)
| tmp-eglibc/sysroots/qemuarm/usr/lib/cmake/Qt5Core/Qt5CoreConfig.cmake:130 (include)
| Tests/RunCMake/CMakeLists.txt:80 (find_package)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
:Release Notes:
Add optional Qt-Tools module.
:Detailed Notes:
Qt provide optional Qt-Tools like "lrelease", etc.
In absence of these tools one can not have i18n and l10n changes automated in build.
Adding "qttools-native" to "meta-qt5" Open Embedded layer provides those tools.
Using Qt-Tools i18n and l10n changes can be auomated.
:Testing Performed:
:QA Notes:
:Issues Addressed:
[GF-11028] Enable build of Qt localization tools in the Qt component
Change-Id: Ieb679249a05501a6ccb25108903a0fd48d4014f0
* fixes:
NOTE: multiple providers are available for runtime qtwebkit-examples (qtwebkit-examples, qtwebkit)
NOTE: consider defining a PREFERRED_PROVIDER entry to match qtwebkit-examples
NOTE: multiple providers are available for runtime qtwebkit-examples-dev (qtwebkit, qtwebkit-examples)
NOTE: consider defining a PREFERRED_PROVIDER entry to match qtwebkit-examples-dev
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* full GL is called desktop in qtbase configure
* without this it will try to autodetect and doesn't fail when
e.g. mesa wasn't built before qtbase
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
They moved the 5.0.2 sources into an archive dir upstream, this updates
the SRC_URI accordingly.
Signed-off-by: Erik Botö <erik.boto@pelagicore.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
This is a osmo-bts that will use libosmo-abis for OML and RSL and
is based on master. In one way it should improve quality in another
it is also testing.
A cmake specific qt5 class which sets all required configuration parameters to
successfully compile cmake based applications.
Signed-off-by: Simon Busch <morphis@gravedo.de>
* only for qt 5.1.0; didn't had the time to port this to 5.0.2
* client recipes need to export OE_QMAKE_PATH_EXTERNAL_HOST_BINS in EXTRA_OECMAKE
Signed-off-by: Simon Busch <morphis@gravedo.de>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* qmake5_paths.bbclass refers to QT_DIR_NAME but it's defined on a higher level which
doesn't makes sense and breaks some use cases
Signed-off-by: Simon Busch <morphis@gravedo.de>
* 5.0.2+git should stay default for people without qt5-versions.inc
* make sure that % isn't appended twice in qt5-versions.inc
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* While building a custom rootfs using poky, meta-oe, meta-ti
and meta-qt5 (all on Dylan branch), I got the following error:
[...]
Computing transaction...error: Can't install
qtbase-tools-5.0.2-r2.0@armv7a_vfp_neon: no package provides /usr/bin/perl
[...]
* syncqt is perl script with /usr/bin/perl shebang, that's why rpm
was complaining (other package managers in OE doesn't check that)
Signed-off-by: Christian Gagneraud <chgans@gna.org>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* this will need small patch in qtmultimedia.pro to run
qtCompileTest(openal) only with OE_OPENAL_ENABLED
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* it's autodetected from sysroot, make it deterministic
* maybe I should allow to change gst-0.10 to 1.0 without
rewriting DEPENDS
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* qtdeclarative was using /usr/lib as HostLibraries causing
WARNING: QA Issue: qtdeclarative: The compile log indicates that host include and/or library paths were used.
Please check the log 'qtdeclarative/5.1.0-r0/temp/log.do_compile' for more information.
DEBUG 1: /OE/oe-core/tmp-eglibc/sysroots/qemux86-64/usr/lib/qt5/mkspecs/features/qt_config.prf:23: QT_MODULE_HOST_LIB_BASE := /usr/lib
* also without this fix qtdeclarative and qtwayland are trying to
build tools against /usr/lib/libQt5Bootstrap.a (without sysroot
prefix)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* qtwayland fails, because we'll need to add qtwayland-native to provide
qtwaylandscanner in QT_EXTERNAL_HOST_BINS location (otherwise build
will try to use /usr/bin/qt5/qtwaylandscanner)
* native recipe is needed to build qtwayland-scanner which
is required after this upstream commit
commit 05d054ad7207764293525661876ebc689940a15c
Author: Samuel R<C3><B8>dal <samuel.rodal@digia.com>
Date: Tue Apr 9 16:52:59 2013 +0200
Introduced qtwaylandscanner.
Signed-off-by: Simon Busch <morphis@gravedo.de>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* some options are tri-state and PACKAGECONFIG doesn't support that
if you need the 3rd option, you need to overwrite PACKAGECONFIG[opt]
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
the patch 0017-Rename-qAbs-Function-for-timeval.patch is also
needed for native else, on Fedora 19, we get :
| ../../include/QtCore/../../../qtbase-opensource-src-5.0.2/src/corelib/global/qglobal.h:483:27: error: from previous declaration 'constexpr T qAbs(const T&) [with T = timeval]'
| Q_DECL_CONSTEXPR inline T qAbs(const T &t) { return t >= 0 ? t : -t; }
Signed-off-by: Eric Bénard <eric@eukrea.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Currently fontconfig support is autodetected. It is better to be
deterministic. Since qtbase-fonts is not installed where fontconfig
will find it, it seems that current users are not using fontconfig.
Signed-off-by: Philip Craig <phil@blackmoth.com.au>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Additional applications might store files in /etc/default and it
is the easiest to backup up all of them. This has the danger that
after an upgrade the default file is not compatible/optimal but
we will need to deal with that once it occurs.
The sysmobts-v2 version was not installed because the file was
not called properly. Rename it to gpsd-default and the sysmobts-v2
version will be picked and installed.
GPSD 3.7 is somehow exiting without us really knowing the cause. The
version 3.9 has been released and we hope that it might just fix the
issues we are experiencing.
This is a bigger update and brings the SAPI queue, readv/writev
and realtime scheduling to the sysmobts software. This will greatly
improve the stability of the software under load.
The proper 'virtual' name for EGL is virtual/egl, not virtual/libegl,
as per meta/recipes-graphics/mesa/mesa-common.inc.
Signed-off-by: Nicolas Dechesne <nicolas.dechesne@linaro.org>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
The previous change modified the defconfig to not include the legacy
pseudo terminal support. Increase the version for the sysmoBTS to have
a kernel without old terminal files.
There is no need to have the various pty files clobber the /dev
directory when they are never used. Do not force a PR upgrade
right now as we don't have a good upgreade strategy for the v1
and the BSC config.
* 0002-Fix-null-pointer-access-in-QQuickVisualDataModelPriv.patch is
applied in newer SRCREV
* 0001-Fix-wrong-calculation-of-viewPort-for-transitions.patch is still
pending
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* 3 backports from newer qtdeclarative:
System Hang while moving App Ribbon
Fix for segfault in some ListView Cases
Make use of BGRA texture format to improve image upload time
* 1 pending fix:
Fix for ViewTransitions not working in horizontal ListViews
Signed-off-by: Florian Haenel <florian.haenel@lge.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
0018 - Offset support for QOpenGLPaintDevice is required for
speedup WebKit hardware rendering.
0019 - Fix FBO restoring in QOpenGLTextureGlyphCache
Jira task
0020 - Uses right texture format to exclude extra convertion.
Signed-off-by: Alexander Irhin <alexander.irhin@lge.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
This way I just need to install sysmocom-openvpn-config and copy
a matching key to the device for the VPN config. The configuration
file is marked as a CONFFILE, this means that modifications to it
will not be lost during upgrade.
- fonts were not installed (at least in qt 5.0.2) so install
them manually.
- split the font package in smaller package as done in qt4 to
save space on the target.
- use OE_QMAKE_PATH_LIBS which corresponds with:
mkspecs/features/qpa/genericunixfontdatabase.prf: fonts.path = $$[QT_INSTALL_LIBS]/fonts
Signed-off-by: Eric Bénard <eric@eukrea.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* WARN: because these modules weren't released yet, we need to make sure to
use SRCREV compatible with selected PV for other modules, old revisions
were compatible with 5.0.[01], I need to test these new SRCREVs to test
git versions
* tested with 5.0.2 and 5.0.2+git%
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* all patches applied in recipes should be also applied in
https://github.com/meta-qt5/qtbase branches to make it
easier to upgrade to new version without loosing any
patches
* apply:
0013-Disable-mkv8snapshot.patch
0014-enables-tslib-device-to-be-read-from-env-variable.patch
0015-qtbase-allow-build-of-examples.patch
also for git version, only
0016-configure-add-tslib-support.patch
is now 5.0.2 specific (similar patch already applied in stable branch)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* maintaining 4 different versions requires a lot more testing of
changes in .inc files and we have only one version of modules without
release (only one _git.bb) and it's impossible to find SRCREV which
would be compatible with all 4 versions.
* 0005-AddSynchoronization-qimagereader.patch was already backported to
5.0.2 in upstream
* 0001-Flickable-Fix-bug-when-flicking-twice-using-touches.patch was
already applied in 5.0.2
* 5.0.2 is now default version (based on DEFAULT_PREFERENCE and in
qt5-versions.inc
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
- the tslib support is not enabled by default and configure doesn't have any option
to active it, so add the option to the build system,
- patch tslib support to get the device from the environment variable
(patch taken from https://github.com/prabindh/qt-configs/blob/master/qt5_1.0_Feb13/tslib.patch
- provide variable to enable tslib support in a bbappend (disabled by default)
- tested on an i.MX5x target with qt 5.0.2
TODO : test with 5.0, 5.0.1 and _git
Signed-off-by: Eric Bénard <eric@eukrea.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
- fix v8 segfault (at least on armv7)
- examples are only provided as sources but in our case, binaries
are more interesting for user to test the features on the target.
Signed-off-by: Eric Bénard <eric@eukrea.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* drop Change-Ids, add Upstream-Status
* 0010-qmake-property-rename-QT_INSTALL_CONFIGURATION-to-QT.patch was
rejected upstream, 0007 was only in meta-qt5/qtbase repo, but wasn't
used in recipes
* 0004-Disable-mkv8snapshot.patch and
0005-AddSynchoronization-qimagereader.patch are special, because they
are applied only to 5.0.0
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* returning QSettings even without existing qtconfig was causing,
e.g. QLibraryInfo::location(QLibraryInfo::PluginsPath) returning
"`pwd`/plugins" instead of correct "/usr/lib/qt5/plugins"
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
We have some images that need the real wget for upgrades, the SSL
root certificate of CA-Cert but that doesn't use task-sysmocom to
build the image. Move these utilities into the tools task.
The rootfs in the UBI is configured to grow to the full size of the
UBI, currently these are 999 LEBs (when there is no kernel volume)
but we had limited the LEBs the filesystem can hold to 400. After
this change the filesystem will grow from the 42MB to more than 120
MB.
I'm not sure if we really want to have a rootfs of that size but we
can at least experiment with it and see if the failure rate is changing
or not.
For the sysmoBTS we want to include a kernel volume inside the UBI as
a fallback in case the mounting of the main kernel is failing. We still
need to modify the boot command of the bootloader to look at this UBI
and to modify the kernel to include the tun/tap modules so the kernel
will be capable of opening a OpenVPN connection.
Currently one needs to write INHERIT += images_types_sysmo into the
conf/local.conf and add ubi-sysmo to the IMAGE_FSTYPES variable.
* and move it to separate .bbclass which is easier to
replace in distro layer when you don't care about
conflicts with qt4
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* endian option was removed from configure
* arch/host-arch option is now deprecated and configure shows
warning when it's used
* both are now autodetected
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* makes it easier to change desktop gl to gles2 in .bbappend
* we cannot explicitly disable opengl with -no-opengl, because then it
fails to build (probably only when mesa headers are found in sysroot)
/qtbase-opensource-src-5.0.0/src/platformsupport/eglconvenience/qeglplatformcontext.cpp:62:72: error: 'QPlatformOpenGLContext' has not been declared
/qtbase-opensource-src-5.0.0/src/platformsupport/eglconvenience/qeglplatformcontext.cpp: In constructor 'QEGLPlatformContext::QEGLPlatformContext(const QSurfaceFormat&, int*, EGLDisplay, EGLenum)':
/qtbase-opensource-src-5.0.0/src/platformsupport/eglconvenience/qeglplatformcontext.cpp:69:70: error: invalid static_cast from type 'int*' to type 'QEGLPlatformContext*'
The same with 5.0.2
Signed-off-by: Simon Busch <morphis@gravedo.de>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* we're using QT_DIR_NAME subdirectory mostly to prevent conflicts with existing files from qt4
but setting it in all OE_QMAKE_PATH_* variables causes all apps which are just using qmake to
build to install e.g. in /qt5 which for stuff like qterminal or something doesn't
sound right (as long as there isn't qterminal4 and qterminal5 recipe)
* some variables are kept with default QT_DIR_NAME, e.g. qml, imports, plugins we can assume that
every application which installs some QML files will install them in location shared by all
* add qt5-native.inc which also adds this QT_DIR_NAME and common
inherits (later will be used also by qtwayland-native.inc)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* "${B}/*" in quotes doesn't match anything, better to remove whole directory
(incluing .files) and recreate it
* not sure why I've added quotes after testing first version, we don't
expect B with spaces.. but I'm a bit scared with rm -rf ${SOME_VAR}
after one glibc upgrade cleaned my whole disk and attached NFS array
when OLD_LOCALE_PATH wasn't detected correctly...
* qmake works well with separate B, use it by default
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* exports are still needed at least for qtbase configure script (which
is using our special eval variant of getQ(X)MakeConf functions
but maybe we should move them only to qtbase now
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* not all modules depend on qtbase, qtbase-native dependency is added
by qmake5.bbclass:
QT5TOOLSDEPENDS ?= "qtbase-native "
DEPENDS_prepend = "${QT5TOOLSDEPENDS}"
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* 5.0.2+git was installing uic in wrong directory STAGING_DIR_NATIVE
was there twice, while debuging this I've noticed that headers and docs
are installed in this wrong directory also in other versions
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* all variables should be now correctly set by qmake
* setting PARALLEL_MAKE in MAKEFLAGS can cause
PARALLEL_MAKE * PARALLEL_MAKE processes, because first
-j is applied on top level directory and then again in
each subdir, but it's faster then make -j PARALLEL_MAKE
only in top directory
* setting QMAKE breaks build in src/tools/bootstrap, because it
forces relative path bin/qmake which isn't correct
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* when uic does not exist in QT_EXTERNAL_HOST_BINS it will silently switch to
/usr/bin/qt5/uic and build will fail a bit later
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* it was working for LIC_FILES_CHKSUM, but not LICENSE
* we need to change order of includes, first version.inc then module.inc
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* MJ: fix PV
* MJ: simplified to use newer classes
* MJ: Add license information
* MJ: use OE_QMAKE_PATH_QML for packaging definition
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* MJ: fix PV
* MJ: simplified to use newer classes
* MJ: Add license information
* MJ: use OE_QMAKE_PATH_QML for packaging definition
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* MJ: simplified to use newer classes
* MJ: Add license information
* MJ: use OE_QMAKE_PATH_QML for packaging definition
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* MJ: simplified to use newer classes
* MJ: Add license information
* MJ: use OE_QMAKE_PATH_QML for packaging definition
* MJ: add missing sqlite3 dependency
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* Disabling this for the time being as it generates invalid
instructions when the target is arm.
Signed-off-by: Mikko Levonmaa <mikko.levonmaa@palm.com>
Signed-off-by: Martin Jansa <martin.jansa@lge.com>
* Fix shader precision definition
* MJ: Fix license information
* MJ: Fix PV
* MJ: add wayland-native to depends for wayland-scaner
* MJ: add libxkbcommon to depends, but make it easy to remove it
Signed-off-by: Mikko Levonmaa <mikko.levonmaa@palm.com>
Signed-off-by: Martin Jansa <martin.jansa@lge.com>
* MJ: Add license information
not sure which parts (or maybe all) is under LICENSE.PREVIEW.COMMERCIAL
* MJ: Fix PV
Signed-off-by: Mikko Levonmaa <mikko.levonmaa@palm.com>
Signed-off-by: Martin Jansa <martin.jansa@lge.com>
* With isis2 touch mocking there is a situation where flickable doesn't
process mouse events even if it has mouse grabbed.
* Patch is cherry-picked from upstream.
Signed-off-by: Heikki Paajanen <heikki.paajanen@palm.com>
Signed-off-by: Martin Jansa <martin.jansa@lge.com>
* thanks to Mikko Levonmaa
* move it from qt5.inc to qmake5_base.bbclass, because it can be useful
for other apps too
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* some modules weren't released yet, so we still need some recipes with
PV = "4.999+5.0.0-beta1+git${SRCPV}"
so having 5.0.2 in .inc filename is a bit misleading
* .inc file still sets default PV (now "5.0.2+git${SRCPV}"), all modules
where SRCREV corresponds to different versions need to set PV inside
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* makes it easier to change them, should be reused also in FILES_*
variables
* table of path variables and their different names available at
https://github.com/meta-qt5/meta-qt5/wiki/Building-with-OE
* all variables have OE_QMAKE_PATH_ prefix and then name from qmake
varaible
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* don't export it, let recipe decide where to call it or even if it
should be called (native recipes are not using it)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* otherwise sysroot from native build can leak to target build
* missing paths like Qml2Imports were defaulting to devault /usr/qml
* synchronize values between qt.conf and configure params
* qt_module.prf replacements were creating -L/usr/lib in .pc files which is bad for cross-compilation
* also use '=' for sysroot in libtool files
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* using ${libdir}/${QT_DIR_NAME} is causing pkgconfig files to be
installed in this prefix too
* modify ArchData variable to move mkspecs files to qt5 prefix (so that
they don't conflict with qt4)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* it overwrites value set from shell env in qmake.conf and ar is loosing cqs params
mkspecs/linux-oe-g++/qmake.conf:QMAKE_AR = $(OE_QMAKE_AR) cqs
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* there are important fixes in qtbase bootstrapping
* create also own directory for patches, easier for many versions to coexist for now
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* patch files are kept in separate dir, because I expect that 5.0.1
will have to live a bit longer for e.g. webkit-qt developers to adapt
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* there are some issues with 5.0.1 and qt-webkit, keep 5.0.0 available
until everybody is ready to upgrade to 5.0.2 or newer
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* add linux-oe-g++ mkspec directly with patch
* add functions to read and eval OE_QMAKE functions from mkspec and
also export them with QMakeVar to be available also for config.tests
* add external-host-bindir parameter to skip building native tools
even when we're in fact cross-compiling (because we have them from
qtbase-native build already).
* use separated ${B} and ${S} and clean ${B} when reconfiguring
stalled qmake cache can be used when configure is reexecuted
cleaning ${B} prevents that and provide cleaner separation
* OE_QMAKE_AR cqs is added by Makefile, having it here too was causing
issues
* isEmpty(QT_EXTERNAL_HOST_BINS) doesn't work, so lets use exist()
even when it allows to incorrectly set wrong directory and build
native tools again (instead of skipping them)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* Upgrade to the master version for a newer (but older than from
what I patched it) jquery
* Queries to js and stylesheet do not appear to add the ?_=NR that
allow caching of requests in chrome
* qmake is trying to call native gdb and we don't depend on gdb-native
(or even provide gdb-native)
* fixes errors like this:
/bin/sh: gdb: command not found
/bin/sh: line 0: test: -gt: unary operator expected
which are not fatal, but still misleading in do_configure output
* g++.conf and linux.conf were outdated a lot, lacking new options like
QMAKE_COMPILER causing a lot of warning like:
Project WARNING: qmake spec does not announce the compiler family. Guessed gcc.
* Instead of replacing whole common files, use custom qmake.conf which
overwrites common setting after including it
* different platform/xplatform can enable cross-compile build in upstream qmake
but that's not the same as what recipes are doing (upstream builds native tools
when bootstraping target qtbase, recipes build native tools with separate
qtbase-native and then want to skip building tools)
* still separate variables for both QMAKESPECs can be useful e.g. for
other native recipes
* quotes are needed, because some people have '-j 9' instead of '-j9'
* this can in theory cause PARALLEL_MAKE x PARALLEL_MAKE threads, because
MAKE itself is using PARALLEL_MAKE threads to run inner makes
* fix paralel build
* default make does good job, cleans mkspecs, installs all headers and
libs needed for qtjsbackend-native
* move native tools to QT_DIR_NAME prefix, this way qt4 and qt5 can be
staged at the same time
* only variables referencing WORKDIR are now QMAKE_PRL_BUILD_DIR
./x86_64-linux/usr/lib/libQt5Network.prl:QMAKE_PRL_BUILD_DIR = /OE/oe-core/tmp-eglibc/work/x86_64-linux/qt5-native/5.0.1-r0.0/qtbase-opensource-src-5.0.1/src/network/
./x86_64-linux/usr/lib/libQt5Xml.prl:QMAKE_PRL_BUILD_DIR = /OE/oe-core/tmp-eglibc/work/x86_64-linux/qt5-native/5.0.1-r0.0/qtbase-opensource-src-5.0.1/src/xml/
./x86_64-linux/usr/lib/libQt5Bootstrap.prl:QMAKE_PRL_BUILD_DIR = /OE/oe-core/tmp-eglibc/work/x86_64-linux/qt5-native/5.0.1-r0.0/qtbase-opensource-src-5.0.1/src/tools/bootstrap/
./x86_64-linux/usr/lib/libQt5Concurrent.prl:QMAKE_PRL_BUILD_DIR = /OE/oe-core/tmp-eglibc/work/x86_64-linux/qt5-native/5.0.1-r0.0/qtbase-opensource-src-5.0.1/src/concurrent/
./x86_64-linux/usr/lib/libQt5Core.prl:QMAKE_PRL_BUILD_DIR = /OE/oe-core/tmp-eglibc/work/x86_64-linux/qt5-native/5.0.1-r0.0/qtbase-opensource-src-5.0.1/src/corelib/
./x86_64-linux/usr/lib/libQt5Test.prl:QMAKE_PRL_BUILD_DIR = /OE/oe-core/tmp-eglibc/work/x86_64-linux/qt5-native/5.0.1-r0.0/qtbase-opensource-src-5.0.1/src/testlib/
./x86_64-linux/usr/lib/libQt5Sql.prl:QMAKE_PRL_BUILD_DIR = /OE/oe-core/tmp-eglibc/work/x86_64-linux/qt5-native/5.0.1-r0.0/qtbase-opensource-src-5.0.1/src/sql/
./x86_64-linux/usr/lib/libQt5DBus.prl:QMAKE_PRL_BUILD_DIR = /OE/oe-core/tmp-eglibc/work/x86_64-linux/qt5-native/5.0.1-r0.0/qtbase-opensource-src-5.0.1/src/dbus/
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* we also had PROVIDES for qt5-tools-native, this makes it a bit easier to see what is what
* drop FILESEXTRAPATHS, not needed after renaming to match BPN with qtbase patches
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
systemd is using the depedencies of the LSB section and according to
the debian documentation all applications accessing /var should depend
on the localfs. This makes the watchdog start properly on systemd systems.
This sanity check is included in the used bbclass directly and to
support both Edison and master we will need to remove the custom
python code due a change in indention in the python code.
We should use perf for future performance problems but it might
require more work (e.g. not to install the source into the -dbg
package, adding newt, etc.). But oprofile is not very nice to
use right now anyway.
Ask the buildsystem to install the manpages to the mandir. This will avoid
the following warning.
WARNING: QA Issue: lmsensors-apps: Files/directories were installed but not shipped
/usr/man
RROR: There is a confusing multiline, partially commented expression on line 22 of file /home/ich/firmen/gesmokom/source/meta-sysmocom-bsp/recipes-extra/gpsd/gpsd_3.7.bb (SRC_URI[md5sum] = "52d9785eaf1a51298bb8900dbde88f98").
Some of our target applications require ext3/ext4 support for mounting
an MMC card. While the sysmobts-v1 hardware does not have an external
MMC card slot it is nice to have the same image available for this
hardware.
Install the sysmobts firmware headers on top of the Osmocom toolchain.
This way the SDK can be used to re-compile both the OpenBSC and the
osmo-bts software.
Make sure the -dev package of the femtobts-api has not runtime
dependencies to a library that does not exist.
The name will be resolved to the hash during parse time. This will
not work when not having access to the internet. Use the sha1 sum
to make offline builds work.
This sadly still prints the firmware reload messages. But this is
necessary right now. The fpgadl/dspdl drivers do not log fatal
failures so the 'success' is the only information we have. This
is to be fixed in the driver.
If the first partition of a MMC card is mounted and has a logs/
directory syslog will start to write log messages to the MMC card
in the directory /media/mmcblk0p1.
This issue is documented in PRJ#72. The easiest is to to keep the
old software running as everything else is racy and start-stop-daemon
does not appear to implement the --retry command. Let's see how
far we get with this workaround.
We want to only log a given loglevel, backport the start up script
from OE-Core to get the LOGLEVEL variable. This removes the MARKINT
and adds DROPDUPLICATES and LOGLEVEL.
* The patch is now included in a temporary branch
* LOGL_NOTICE is now used instead og the LOGL_INFO level which will
be ignored by default on sysmocom systems.
The kernel has been rebased to target 3.2.40 with various bug
fixes in the upstream work. The sysmobts v2 config has been
changed to allocate space for two more UARTs. These are required
for the multi-trx configuration and they make a warning go away.
* Fix the link timeout handling as found in post-review.
* Fix the set bts attr parsing due a funny parsing error.
* Sleep 10 seconds after the execution, reload DSP and FPGA firmware
We could use "fakeroot do_install() {}" but this will still not
make the spool directory be owned by uucp. Use a postinst script
to make this up. The logic is copied from the cronie bitbake file
and we will see how this works at image creation time.
Adjust the oom_score_adj of the shell script to avoid ever beeing
killed. Adjust the score of the child and nice it. Sleep a given
interval before respawning it.
This script creates a .pid file so it can be used by a sysv init
script to stop the respawner and the client.
Use the "() &" syntax to adjust the score before executing the new
command. This is removing the reace from the previous commit. Update
the PR to force rebuilding new packages.
Adjust the score for the monitor process. The issue is that the
score is inherited by the process. Right now the score of the
child is adjusted _after_ the process has been spawned. This is
obviously a race condition.
This is using some brute force. Make the shell script write a PID
file and on stop kill the respawn script and then all process names
that are spawned. For the stop we need to know that this is a bash
script.
FIXME: this will actually assume a GPS that is indirectly attached
to a GSM Modem which will be handled by openmoko-gsmd and forwarded via
UDP. However, we might have a GPS receiver attached directly to the
GPS UART in other hardware versions!
This will include
* Fix of CHAN_NR IEI in common channel RSL messages generated by BTS
* zecke/request_queueing
* Improve shutdown of DSP on exit
* various calib updates
* GSMTAP: use RxLevel and RxQual
* DSP power control bypass on ul_power_target==0
* Add radio link timeout procedure
They will cause the building of cmake-native to fail as it
will parse the output of qmake -query and is not expecting
to find the added lines in the output
Signed-off-by: Mikko Levonmaa <mikko.levonmaa@gmail.com>
The upgrade of the unit broke due killing all client connections. opkg
got a SIGHUP and didn't write out the state files. Do not kill the
client connections on upgrade.
The modules depend on kernel-base (renamed kernel-3.2.35) package,
the tasks depend on some modules and nothing needed the kernel package.
In OE Core/Poky the machines depend on kernel-modules which will pull
the kernel package into the rootfs. We do it by hand using the essential
depends and bumping the PR for the task-core-boot.
I have verified that on a kernel ugrade task-core-boot will be updated
and the kernel package is installed.
The previous patch made sure a truncated key would be re-generated
but it broke the case that there is no key at all. This code is
handling both cases now. Use quotes as code outside this method
is using them too. Use the -a and not the && operator to work with
the busybox version of [ and ].
The latest release currently does not support the RevC hardware. Install
both 3.0.1pre and 3.1 (luckily ubifs compresses them) and set a symlink.
Use the EEPROM size as an indicator if we are on revD or revC hardware
and then set the symlink. It has been tested for revC and needs to be tested
on revD hardware.
* and change DEFAULT_PREFERRENCE for git recipes negative again
* bump SRCREV in git recipes to latest and update PV
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Re-work the way we distribute the header and firmware files. The header
files may be accessed by everyone but the firmware may only be used by
owners of our device. Create a header only repository with all the header
versions of the firmware and create tarballs for specific releases that
will be provided to our customers. Use COMPATIBLE_MACHINE to pick the right
firmware and API version depending on sysmoBTSv1 and sysmoBTSv2.
This has been fixed in upstream of OE-Core and in newer versions
of Yocto. For Edison provide our own version of the pkg_postinst
and pkg_postrm routines. It is not clear yet if this conflicts with
newer versions of poky but all that could happen is that the symlink
is changed twice and removed twice but the command ends with a || true
so this should not cause any kind of fault.
Created new variables OE_CROSS_INSTALL_DATA and OE_CROSS_HOST_DATA
to support the installation and configuration of qtbase and dependant
modules.
Added paths for Qml2Imports
Signed-off-by: Mikko Levonmaa <mikko.levonmaa@gmail.com>
This allows the subsequent modules to install correctly. Also during
do_configure we stage target specific file into the staging area and
modify some prf files inorder to make the module (qt_lib_<module>.pri)
files install correctly
Signed-off-by: Mikko Levonmaa <mikko.levonmaa@gmail.com>
They are placed under STAGING_DATADIR as they need to be
kept separate from the native side. The reason for doing so
is that some qt modules require native tools and the mkspecs
in STAGING_DATADIR_NATIVE cannot be polluted with the target
mkspecs
There are still some packaging issues
Signed-off-by: Mikko Levonmaa <mikko.levonmaa@gmail.com>
* otherwise we need for -system-xcb
libxcb-native libxau-native libxdmcp-native xcb-util-wm-native
xcb-util-image-native xcb-util-keysyms-native
or for -qt-xcb
libxcb-native libxau-native libxdmcp-native
and fix underlinking in xcb linking test (QMAKE_LIBS_XCB)
| linking xcb
| /usr/lib/gcc/x86_64-pc-linux-gnu/4.7.2/../../../../x86_64-pc-linux-gnu/bin/ld: warning: libXau.so.6, needed by /OE/oe-core/tmp-eglibc/sysroots/x86_64-linux/usr/lib/libxcb.so, not found (try using -rpath or -rpath-link)
| /usr/lib/gcc/x86_64-pc-linux-gnu/4.7.2/../../../../x86_64-pc-linux-gnu/bin/ld: warning: libXdmcp.so.6, needed by /OE/oe-core/tmp-eglibc/sysroots/x86_64-linux/usr/lib/libxcb.so, not found (try using -rpath or -rpath-link)
| /OE/oe-core/tmp-eglibc/sysroots/x86_64-linux/usr/lib/libxcb.so: undefined reference to `XauDisposeAuth'
| /OE/oe-core/tmp-eglibc/sysroots/x86_64-linux/usr/lib/libxcb.so: undefined reference to `XauGetBestAuthByAddr'
| /OE/oe-core/tmp-eglibc/sysroots/x86_64-linux/usr/lib/libxcb.so: undefined reference to `XdmcpWrap'
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
The pri files get correctly placed under the work dir, but fail
to stage correctly. Also there are some packaging issues
Signed-off-by: Mikko Levonmaa <mikko.levonmaa@gmail.com>
This patch has been cleaned to the minimum version and is ready for
sending to upstream. Once it is accepted, a backport of the final
version will be included here.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
* also add version specific .inc files and move all version specific
variables to it
* upgrade 5.0.0-beta1 to 5.0.0-beta2 and add include for latest git
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
only the first prototype board (on the v2B mainboard) was using ttyS0, messing
up console access. This has been fixed in v2.x of the db9 daughterboard for
the v2C mainboards.
Disabled tests and exmaples in the native build. Also do_install
now installs the qmake host binary
Signed-off-by: Mikko Levonmaa <mikko.levonmaa@gmail.com>
The idea is that the basic extension resides inside the the normal
recipe directories. Depending on the version of poky the bbappends
need to be included. Appends applying to all versions will be put
into the shared folder, specific ones into the specific directory.
Once a bbappend can not be shared anymore it needs to be copied.
Make the start/stop of the NITB, BSC and MGCP HW work, remove the
force-reload option as it didn't do anything and omit writing counters
into the database.
META_TELEPHONY_OSMO_INC functions as kind of a reverse dependency,
packages using libosmocore will now rebuild and link to the newer
soversion of the libraries. Upgrade OpenBSC due an ABI change in
the GPRS/GB code.
The UBL allows to interrupt the NAND boot mode and switch to the
UART mode for recovery. Everyone with physical access will be able
to interrupt the boot.
Install the sysmobts calibration utility. task-sysmocom-bts is
installed on every bts and will now pull in the calibration utility.
The sysmobts screen needs to be stopped before this application
can run.
First ca-certificates does not create a symlink from /etc/ssl/certs
to /usr/lib/ssl/certs, second it does not run the c_rehash perl utility
making the recipe pretty useless and third it trusts way too many
cert authorities.
* Enable verbose detection, add CP2012 based serial adapter support
* Only enable USB1.1 HCD for now to fix the detection of the USB serial
adapters
* Enable more filesystems in the config to allow upgrading the boot
partition during the runtime (after installing the right codepage)
socat is useful for the self-hosted-image work.
The original recipe is from OE:
http://git.openembedded.org/openembedded/tree/recipes/socat/
and I upgraded it to 1.7.2.0.
Thank Koen Kooi <koen@dominion.thruhere.net> for suggesting
how to assign the LICENSE field with a proper value.
(From OE-Core rev: b1771ff0ad39250678bd53b0ae7543c9365572f5)
Signed-off-by: Dexuan Cui <dexuan.cui@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
RevB and RevC feature different kind of nand flash. RevB is still
pre-production hardware so for the few units allow to set the right
flash parameters in the local.conf.
tcpdump can create a new log file every couple of bytes, it can also
be asked to limit the number of logfiles and wraps the counter back to
zero, and at least tcpdump 4.1 can also compress the log file with an
external script. Create a start script around this feature to log to
a given directory but not use more space than allowed to. The script
was created by Jan Luebbe and then modified.
This might be a general TLS issue with Asterisk on ARM. I have not
looked into why this is happening. Create a band-aid.
Crash:
(gdb) bt
it crashed when formating/printing the &addr in the rtcp recv method
...
RTCP:
Packet generated from Ekiga 3.2.7 (Debian Unstable).
Cause:
Not very clear, maybe something with TLS handling of asterisk
DESCRIPTION="Cinematic Experience collects many of the new Qt5 QtQuick 2.0 features into the same UX demo application. It uses particles, sprites, path animation, custom shaders etc. features which Qt5 introduces for QML UIs."