I figured out.
udevadm monitor ist provided by the SystemV package.
Problem solved!
Stefan
ESA Elektroschaltanlagen Grimma GmbH
Broner Ring 30
04668 Grimma
Telefon: +49 3437 9211 181
Telefax: +49 3437 9211 26
E-Mail:
Hej,
Actually I am playing a bit arround with udev rules. For that I like to
use the udevadm monitor. The udev is provided by
/meta/recipes-core/udev/eudev_3.2.2.bb
In this recipe there is no udevadm option. While checking, I saw it was
once in udev-utils and been moved to udev.inc.
Hej ho
my Workflow for adjusting uboot (u-boot-at91):
1.) build the original uboot
2.) apply changes
3.) create patch
3.1.) if you reconfigure it -> save/copy the config
3.2.) if you adding code -> use meld or something similar to create a
patch
4.) create u-boot-at91_%.bbappend
4.1) copy
Hej
It's friday and I have some time to play a bit around. I like to build a
simple test-adapter using a RPi. I want to connect some of the 40GPIO pins
with a custom mainboard. For that I like to define the pin behaviour by an
own dts file.
I created my own dts file (testboard.dts) + a simple
Hej
I like to setup a central host for the packages build with yocto. The goal
is, that everyone can download the "new" packages from a server in the
local network. I am looking for a description how to archive that or some
tips to start from.
Regards!
Stefan
Hej
I like to configure syslog. It seems that are more than one way to set up
syslog. I am asking for the comon way/steps to do that.
I have several use cases. To simplify I like to ask how to configure
syslog to write an infinity long log file in /var/log/.
Following steps:
1.) configure
Hej
I want to create an own meta-bsp package for my custom mainboard based on
an ATMEL SAMA5D35 SoC. For that I checked the documentation
http://www.yoctoproject.org/docs/2.2/bsp-guide/bsp-guide.html#using-the-yocto-projects-bsp-tools
.
For me it is not clear how to create a bsp package based
Hej
I want to install apache2 + a website. For that I need to copy my website
to the std. dir defined in the apache config file. In my thinking my
website uses the apache service to be aviable and therefore is an own
package (myweb.bb). This package contains the dependency to the "apache2"
When I run 'bitbake core-image-minimal' before it works. Think Roberts
work arround points at the right spot(because the populate sysroot is
called through core-image-minimal build). Maybe add this trick to the
documentation.
Stefan
Hej
Some error occurred. I give a print (sorry for that German text - but the
error should be clear):
##
GEN ./Makefile
HOSTLD scripts/kconfig/mconf
scripts/kconfig/mconf.o: In Funktion `show_help':
mconf.c:(.text+0x914): Nicht definierter Verweis auf `stdscr'
Thanks for the help! The solution is simple. As mentioned by Ross No
do_fetch extension is needed. The extract + copy goes into do_install.
part of the recipe
do_install () {
install -d ${D}/opt
tar -xzf example.tar.gz -C ${D}/opt/
}
FILES_${PN} += "/opt/*"
Hej
I like to build an package out of a git repro. This git repro contains
compressed files. One of these files I like to copy into my package. I
looked into the documentation, but it is only described how to use the
fetcher of bitbake. Think I need to add some extra lines of instruction
Hej ho
I have an app that is using syslog. For it I like to set the amount of log
data.
I am using the std. yocto(krogoth) together with the std. meta-atmel
layer. This layers are installing the "busybox-syslog"-package. This comes
with the "/etc/syslog.conf" file.
My question is: How to
Hej
Some software can be only tested on the final system. For that it would be
usefull to start with a debug version of that app and change it later to
release.
In my case I have a software which do logging. The amount of the logs
depends on type of the build (Release skips all debug
Hej
If have a simple problem. On one system a program is started as a daemon
and on the other one not. In my recipe the program is build. by using
update-rc.d I create the init script for systemV. The code looks like:
##
SUMMARY = "myApp"
SECTION = "myApp"
LICENSE = "CLOSED"
DEPENDS =
The problem was fixed. Updating meta-atmel to the current version fixes
the dpkg error by appending the architecture parameter to the dkpg recipe
(
https://github.com/linux4sam/meta-atmel/commit/63d4287f50b89496a4c72e3b5845e59c55887451
).
Regards!
Stefan Jaritz
Hej ho
Thank you. I already know the files. My question was targeting the way how
to set the tune option manually.
Maybee something is missed up by the layers. My bblayers.conf looks like:
#
# POKY_BBLAYERS_CONF_VERSION is increased each time
build/conf/bblayers.conf
# changes
when forcing dpkg it is installed and it works.
Stefan Jaritz
ESA Elektroschaltanlagen Grimma GmbH
Broner Ring 30
04668 Grimma
Telefon: +49 3437 9211 176
Telefax: +49 3437 9211 26
E-Mail: s.jar...@esa-grimma.de
Internet:
I read the printings like you and I am confused.
let's take valgrind as example:
1.) the rootfs build goes to:
/poky/build/tmp/work/cortexa5hf-vfp-poky-linux-gnueabi/valgrind/3.11.0-r0/deploy-debs/cortexa5hf-vfp/valgrind_3.11.0-r0_armhf.deb
the extension of the deb package say it its hard float.
hej
@find out:
transfer a build package into the evalboard and try to install it. I am
using deb packages. dpkg gives me an error because of architecture.
@print configuration output:
###
Build Configuration:
BB_VERSION= "1.30.0"
BUILD_SYS = "x86_64-linux"
Hej
I am building for the Atmel Evalbaord SAMA5D3x a Yocto Image. It works
fine, but sysroot toolchain is set to soft float instead of hard float.
How can I set the sysroot toolchain to hard float?
My actual configuration:
1.) use meta-atmel layer
2.) in /build/conf/layer.conf -> MACHINE ??=
Hej
I am using the actual Yocto version. A Yocto Linux I is build based on the
meta-atmel layer. The system seems to be "armel" but the packages are for
"armhf". When running bitbake it shows:
##
Build Configuration:
BB_VERSION= "1.30.0"
BUILD_SYS = "x86_64-linux"
Hej
It is Friday and the week is almost over - it is time for some image
beautifications!
Is there a switch to activate the console colours by default? If not -
what is the preferred way to produce a nice coloured console?
Regards!
Stefan Jaritz
It works fine! Problem solved
Stefan Jaritz
ESA Elektroschaltanlagen Grimma GmbH
Broner Ring 30
04668 Grimma
Telefon: +49 3437 9211 176
Telefax: +49 3437 9211 26
E-Mail: s.jar...@esa-grimma.de
Internet: www.esa-grimma.de
Hej
I want to create a recipe for an application. This application is
generated through cmake. In the cmake file "CMakeLists.txt" is the
passage:
##
include(FindPythonInterp)
include(FindPythonLibs)
add_custom_command(OUTPUT "${CMAKE_BINARY_DIR}/error.cpp"
Yes - The installation of the can_if into /etc/init.d was missing.
By adding:
do_install () {
install -d ${D}${sysconfdir}/init.d
cat ${WORKDIR}/can_if | \
sed -e 's,/etc,${sysconfdir},g' \
-e 's,/usr/sbin,${sbindir},g' \
-e
Hej
I run into a problem when I tried to install 3 initscrips out of one
recipe. My recipe looks like:
#
SUMMARY = "demo cfg"
SECTION = "demo"
LICENSE = "CLOSED"
inherit update-rc.d
RDEPENDS_${PN} = "bash initscripts"
SRC_URI = "file://startA \
file://startB \
Hej
I want to install my software packages to the usr/local/bin dir. That is
why I need to change the prefix form usr to usr/local. How to archive
that?
The software is build through a cmake. Is there somethingt I need to think
extra for setting the prefix?
Maybe I found some inconsistency:
Problem solved! Thank you for the help.
It was solved by:
do_install () {
...
install -d ${D}/media/sd1
...
}
FILES_${PN} += "media/sd1"
@Mike:
I am using the meta-atmel. The mount dir is created. In my case I have a
rule which mounts the sd card to mount/sd1, That's why I need to create
Hej
I managed to create and install a rule that should mount a sd card to
"/media/sd1". To finish it, I need to create the directory "sd1" in
media.
My recipe for the rule looks like:
###
SUMMARY = "the udev rules for the board"
SECTION = "rules"
LICENSE = "CLOSED"
SRC_URI =
Hej,
I want to start a service that generates Sockets for the CAN Modules.
Manually configuring the system is no problem, but I like to have it done
by yocto. Below I give the code of my recipe (socketcan.bb):
#
SUMMARY = "the config for the can socket interface"
SECTION
Hej
I have several recipes which are based on cmake. At cmake is the "make
package" enabled. It is based on CPack(debian).
I am using the predefined GNU installation dirs by
"include(GNUInstallDirs)" and its variables like CMAKE_INSTALL_FULL_BINDIR
etc.
Hej
I want to build a library. It comes with a cmake based build. In these
process the library is build and deployed together with 2 links. Like
that:
libMyLib:
usr/lib/test/mylib.so
usr/lib/test/mylib.so.0
usr/lib/test/mylib.so.1.0.0
mylib.so and mylib.so.0 are links to mylib.so.1.0.0
The
Hej Daniel,
Building the modules as rpm works fine but it fails for deb. The deb
generation process seems a little buggy. I have discovered problems with
naming recipes (the QA tries to find the license by a "lowercased" link).
I checked the package dir. It is fine. The 3 files (a.ko, b.ko, &
Hej
I have a problem when running do_rootfs() with some own kernel drivers
included.
My kernel recipe builds some easy GPIO drivers.
# They come as one package:
myKernelModules:
-> a.ko
-> b.ko
-> c.ko
# I build a recipe which creates:
myKernelModules.bb
->
Hej
I have encountered a problem, when I switched from rpm to deb. I have a
recipe which generates 3 kernel modules. The compiling and packaging works
fine. But when I do the do_rootfs it fails with:
##
ERROR: core-image-minimal-1.0-r0 do_rootfs: Unable to install packages.
Hej
I found the error at the QA check of the debian packages. The deb QA modul
has problems resolving the links into the
"sysroots//pkgdata/runtime-resolve". This is because there is a
lowcase conversation of the recipe name:
p.e.:
"helloMyWorld.bb"
the name of the link is "helloMyWorld"
Hej
I deleted the tmp and rebuild it. But the error stays. I added a licence
file to the repro and added the
LICENSE="ESA"
LIC_FILES_CHKSUM="files://ESAlicense.txt;md5=.."
What's the connection between the LICENSE and LIC_FILES_CHKSUM field?
How the license manifest is build?
I think the
Hej
Thanks for the infos. I figured out that the common tutorial under
meta-skeleton does not covers the change of the root password. Although
"usermod -P" does not exits. Now the "usermod -p" wants a encrypted
password. Looks like most of the examples at the www are little outdated.
My
Hej
I switched from RPM to DEB because the board configuration should be
handled by a package manager. By doing so I ran into a QA problem. All the
software provided by my colleagues I maked with
LICENSE = "CLOSED"
because they do not provide a license file. This works fine with the rpm
Thanks & done in that way
but the bitbake error "core-image-minimal-1.0-r0 do_rootfs:
myNewPassword.bb not found in the base feeds" still remains
Stefan Jaritz
ESA Elektroschaltanlagen Grimma GmbH
Broner Ring 30
04668 Grimma
Telefon:
Hej
I want to change the root password. For that I create the recipe
"myNewPassword.bb" with:
SUMMARY = "my new password"
SECTION = "new"
LICENSE = "CLOSED"
inherit extrausers
EXTRA_USERS_PARAMS = "\
usermod -P xyz root; \
"
compiling and packaging
Hej
I have a git source which need some submodules. For that I need to execute
"git submodule init" and "git submodule update". Any idea how to do it in
a comon way?
Regards!
Stefan Jaritz
ESA Elektroschaltanlagen Grimma GmbH
Hej
Problem solved! It worked like described by Rebert and on Stack Overflow.
Thanks for the help
Stefan Jaritz
ESA Elektroschaltanlagen Grimma GmbH
Broner Ring 30
04668 Grimma
Telefon: +49 3437 9211 176
Telefax: +49 3437 9211 26
Thanks for the reply
Like you thought - I want to include the lib + headers.
I included the "myLib" via
IMAGE_INSTALL_append = " myLib"
and have a project
IMAGE_INSTALL_append = " myProject"
which has in it's recipe
DEPENDS = "myLib"
In my eyes this should be enough.
When looking into
Hej,
I am looking for an example where is shown how to include an own library
into the toolchain.
Let's call the library "myLib". For it I created an recipe "myLib.bb" and
added:
BBCLASSEXTEND = "native nativesdk"
In my "local.conf" I added:
IMAGE_INSTALL_append = " myLib"
When
Hej
I want to build a recipe for a cmake based application. To make it run I
need to set a cached cmake variable.
What is the common/best way to achive it?
Regards!
Stefan Jaritz
ESA Elektroschaltanlagen Grimma GmbH
Broner Ring 30
Hej
I a cmake based application, some code generation via dbusxx-xml2cpp. It
is provided by the dbuss-c++ lib. A recipe can be found under:
http://git.yoctoproject.org/cgit/cgit.cgi/meta-ivi/tree/meta-ivi-demo/recipes-extended/dbus-c++?id=b90c2a20744e1e82b2169a88e68d7677bec907b6
It working and
Hej
I have some general question. Actual I am try to build an older software.
It is based on cmake. At the build process doxygen, python, sed,
dbusxx-xml2cpp are called. They are included via the
"findprogram"/"find_package" command.
How to include external programs like python, sed etc. in
Hej
I try to build the libdbus-c++ libary. The build process is based on
autotools. When compiling it do_configure throws an error. The macro
"AC_DEFINE" is not found. Any idea how to make autoreconf run?
Below I listed to recipe and after the Error plot
# libdbus-c++_0.9.bb
I solved that problem
Using
IMAGE_INSTALL += "ledmodule"
instead of
IMAGE_INSTALL_append = " ledmodule"
caused that the module was installed.
I found the hint in
https://communities.intel.com/thread/56810?start=0=0
Thanks for the help. Maybe a tip in the recipe template "hello_mod" could
Thanks for taking the time to help me
I tried your approach but it does not worked.
I tell what I figured out so far:
I.) checking the target system
1.) When flashing the images into my device and boot the system, the
driver is not loaded.
2.) lsmod shows that there is no driver loaded
3.)
Hej
I created a recipe for a own modul. It's similar to the example given at:
http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta-skeleton/recipes-kernel/hello-mod
I checked the bitbake process. A "kernel-module-" package is
created and filled with the ko-file.
Actually I am failing to
Hej
I like to build a module. I am working on a SAMA5D3x Evalboard form Atmel.
For that I use the meta-atmel recipes.
My driver should toggle a pin. The pins are defined in the "mach/gpio.h".
Any ideas how to configure that the atmel linux is creating the includes?
Maybe I forgot to set some
Thanks so far!
I have 3 questions for my understanding:
1.) How to define extra Include Paths if the module needs some extra
Includes?
Like the "mach/gpio.h", which should be under
"/arch/arm/mach-at91/include" and is needed for the PIN
definitions.
Should it be done at the
Hej,
I want to build a recipe for a simple GPIO Module. I found no tutorial at
the documentation. What is the comon way to build a simple kernel module?
The Makefile looks like:
#
obj-m += ledmodule.o
SRC := $(shell pwd)
all: modules
modules:
$(MAKE) -C $(KERNEL_SRC) M=$(SRC)
Thanks,
What is when the HG comes over SSH with authentication? Like:
SRC_URI = "hg://myHG/ABC;module="dev";protocol=ssh;rev=42"
with user for example "yocto"
I think the syntax could be: SRC_URI =
"hg://yocto@myHG/ABC;module="dev";protocol=ssh;rev=42"
Regards
Stefan Jaritz
Entwickler
Hej,
Actually I try to build a recipe for a Mercurial SCM. The hg sources are
local. In my bb File I have this lines:
SRCREV = "486c7001d4e5"
SRC_URI = "hg://myHG"
I got the error:
ERROR: Logfile of failure stored in:
Hej
Finally we made it - Thanks to you all for helping me! I attached my bb
file for that case someone is searching for a bitbake recipe for building
the libhiredis
##
SUMMARY = "libhiredis"
SECTION = "lib"
LICENSE = "MIT"
LIC_FILES_CHKSUM =
Hej
sorry - that does not work. Maybe you can show me some working/common
code.
Do I have to create the FILE_${PV} or/and FILE_${PV}-dev,
FILE_${PV}-static?
The builded files go into the Git dir & the autotools-brokensep is
generating the package dir. I repeat my question: What are the
Hej
The makefile has an install section. I relocated it with the ${D}. By
doing that I managed that the compiler filled the "package" directory. But
there is still a packing issue.
Below is the log (and below of it is the libhiredis.bb).
Now my question: How to define the FILES variable for a
Hej
I like to include the hiredis library. I created a recipe and modified the
EXTRA_OEMAKE. The compilation works (in the git directory object, "so" and
"a" files are generated). But the generated files are not placed into
packages.
Any ideas?
below there is the "libhiredis.bb" file:
Hej,
I am new to yocto. I like to create an recipe for the g3log library. I
managed to get the git sources and start the cmake compile. But it stops
when it tries to compile "#include ". Do you have an idea how to
fix that?
Below there is the content of the "g3log.bb":
Hej,
I read that there is a way to the binary configuration (
https://wiki.yoctoproject.org/wiki/Binary_configuration_support). It is
described jow this internally works. But I am missing an example how to
use it praticaly.
My learning system is a SAMA5D3x Eval board from Atmel. For that I am
64 matches
Mail list logo
