Compiling linux for embedded system using Arch.

Hi. I build the vanilla kernel for an embedded system using special buildsystem. This buildsystem calls
make INSTALL_MOD_PATH=embedded_fs_path modules_install
The kernel's Makefile installs the modules to embedded_fs_path/lib/modules/$(KERNELRELEASE) and calls /sbin/depmod -b embedded_fs_path $(KERNELRELEASE).
This operation fails under Archlinux because /sbin/depmod is a part of kmod and it expects that modules are placed to /usr/lib/ instead of /lib, so it tryes to make dependencies in embedded_fs_path/usr/lib/modules/$(KERNELRELEASE) instead of embedded_fs_path/lib/modules/$(KERNELRELEASE).
What is correct solution for this issue? I can pass DEPMOD variable to kernel's Makefile to use something else instead of /sbin/depmod, but i don't know value of this variable should be.
Is there any command line options for kmod to force using /lib instead of /usr/lib? As I've seen in kmod sources, there aren't.
Is there any package in repositiries or AUR or somewhere else which provides an application that behaves the same way as old depmod with /lib/ modules location?
Thanks.

tavianator wrote:Fedora Core 7? Wow, that's ambitious. I'd suggest building in a virtual machine or chroot with the target system installed.
centos is probably worse. They're still on 2.6.18 as far as i can tell.
I wanted to avoid chroot/virtual machine (ran one of those for a while). The reason is that my current development build is the same as what is delivered. I don't want to lose that....having to deal with different compilers, libraries, etc can be a pain.
I'll also look into whether or not I can install a "glibc-compat" type package that could alleviate the loader.
Here's the telling part:
the new exectuble:
$ file <executable>
<executable>: ELF 64-bit LSB executable, AMD x86-64, version 1 (SYSV), for GNU/Linux 2.6.27, dynamically linked (uses shared libs), for GNU/Linux 2.6.27
the old executable:
$ file <executable>
<executable>: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.8
Okay....issue seems to be in the glibc PKGBUILD.
it has the line:
--enable-kernel=2.6.27 \
passed to glibc configuration.
Causes big time problems for creating compatible packages.
Last edited by bnolsen (2011-05-05 21:27:20)

Similar Messages

JAVA for embedded systems: just a myth ?

Hi,
I'm relatively new to JAVA (a year or so), and so far I'm fascinated. I've only developed for PC using J2SE, although I'heard a lot of hipe about J2ME, which I guess is the JAVA for embedded systems (correct me if I'm wrong here).
I worked a lot with microcontrollers, and I think it would be really nice to program them using JAVA. I googled serched for JAVA enabled embedded systems, but I couldn't find anything.
Anyway, this morning I got to this site and find a very interesting article, which I'd like to share with you and hear your opinions about the 5th myth.
The link is: http://www.chipcenter.com/knowledge_centers/embedded/todays_feature/showArticle.jhtml?articleID=10300909
Cheers!!

>... very interesting article, which I'd like to share with
you and hear your opinions about the 5th myth.
Very interesting. I'd always wondered what happened to the Java chips...
Here's a thought, and an associated question, if any of you reading this happen to be up on modern processor architecture...
What about microcode? Back in school (decades ago) I learned a bit about microcode, which was like very simple, low-level instructions that a processor would "execute" to execute each assembly/machine language instruction. The idea was that if you needed to tweak the instruction set of a processor, you could rewrite its microcode. Seems like this would lend itself nicely to implementing a JVM in "hardware"... The question is, do they even use microcode anymore? With the pursuit of speed and optimization, I wonder if they didn't abandon it and go with highly specialized/optimized circuits instead...
Back to the original question -- it would be somewhat ironic if Java weren't available for embedding systems. The original target system was set-top boxes, etc. I've tried a few JVMs on my Palm -- they're getting there!

Crypto libraries for embedded systems

Hi, I need some crypto libraries for my Xlet running on a MHP receiver. I need that these libraries can use the public key extracted from a certificate and encrypt a digest using that public key. I can't use javax.crypto because it doesn't work on a limited MHP 1.0.2 receiver. I need something studied for it. Thank you!!!
Edited by: fabio80s on Oct 23, 2009 3:24 AM

>... very interesting article, which I'd like to share with
you and hear your opinions about the 5th myth.
Very interesting. I'd always wondered what happened to the Java chips...
Here's a thought, and an associated question, if any of you reading this happen to be up on modern processor architecture...
What about microcode? Back in school (decades ago) I learned a bit about microcode, which was like very simple, low-level instructions that a processor would "execute" to execute each assembly/machine language instruction. The idea was that if you needed to tweak the instruction set of a processor, you could rewrite its microcode. Seems like this would lend itself nicely to implementing a JVM in "hardware"... The question is, do they even use microcode anymore? With the pursuit of speed and optimization, I wonder if they didn't abandon it and go with highly specialized/optimized circuits instead...
Back to the original question -- it would be somewhat ironic if Java weren't available for embedding systems. The original target system was set-top boxes, etc. I've tried a few JVMs on my Palm -- they're getting there!

Bonjour for embedded system

Hi,
I have followed the steps provided in the install file to intilize the bonjour for embedded system . first I generated the shared ib for bonjour source code for my embedded system and then I initilized it by calling mDNS_Init api.
Since I just need to browse at my client , I called the browsing fucntion as per dns_sd.h , but I got the below error .
mDNS_StartBrowse: mDNSPlatformRawTime went backwards by 1669225997 ticks; setting correction factor to 1669225997
11-17 00:28:08.481: D/bonjour(1280): mDNS_StartBrowse: mDNS_Unlock: ERROR! m->timenow aready zero
11-17 00:28:08.481: D/bonjour(1280): DNSServiceBrowse("_appl._tcp.", "local.") failed: mDNS_StartBrowse (-65546)
11-17 00:28:08.490: I/System.out(1280): DNSService call failed -65546
11-17 00:32:24.720: E/ThrottleService(161): problem during onPollAlarm: java.lang.IllegalStateException: problem parsing stats:
I hope apple forum will help for this problem ..

I am able to fix the above issue , but subtype search on the embedded systems throws an erros
ex if type="_book._tcp.,author" throws bad param error but same thing works on the linux pc . Can some tell me how to pass the subtypes in embedded application

Is it possible to create a program for embedded system

Is it possible to create a program for embedded system?

Yes. Many solutions.
1 = http://jcx.systronix.com/
2 = http://www.harbaum.org/till/nanovm/index.shtml (see links to Asuro)
And many more but I'm too lazy too look them for you ...
Dan

Do many FPGA/embedded developers use Arch?

I was wondering, since I'm a developer and considering Arch for my workstation, are there many FPGA, embedded etc. designers/developers in the Arch community or at least using Arch for their work?
I don't know, if Arch is as popular among software developers of many kinds (or should I say languages), as i.e. Gentoo is, but looking at the whole distro concept (minimalism, simplicity, configurability etc.) it should draw a lot them.
So.. it would be good to know, that, besides Arch/Linux specific support, help for that kind of development speciallity is also not far away

Anthony Bentley wrote:
In a general Unix sense, the biggest advantage over Windows for embedded development is definitely package management.
At a previous job, I wrote code for some Olimex ARM boards, to help a grad student with his thesis. He had set up a development environment on his home Windows PC, but later could not reproduce the installation on the lab computers: finding a cross‐compiled Windows ARM toolchain, compiling OpenOCD in Cygwin, and then configuring Eclipse to tie it all together.
On my OpenBSD laptop, I did “pkg_add openocd arm-elf-gcc”, wrote a simple Hello World in C, and got the board blinking an LED in no time at all. He was impressed.
So a great and simple way to make your distro more usable for embedded development is to create packages when they don’t exist. It’s pragmatic (because really, nobody wants to remember the right configure flags and recompile GCC again). Arch seems to have packages for quite a few embedded toolchains and/or communication software, which is nice.
I've packaged systemc to arch. I would like to have it on AUR. Thou I dont have the will to write all the flags of the package. I have something which works perfectly to my needs. I would happily hand the code if someone would like to place it properly in AUR.

LabView for Embedded Systems

Hi all.
I intend to develop an Embedded System, based on an ARM9 MCU, and I´ve been wondering if LabView could be used to create stand alone applications to run on such a system.
I can use Windows CE, or Linux on my system, but I prefer Linux.
Can I develop applications directly on LabView for Linux?
What minimal resources would be needed to run an average LabView application on such a system, like memory capacity, minimal MCU clock speed, etc?
I´ll apretiate any comments, since I have no ideia about what can be done with LabView out
of the PC universe.
Thanks alot.

You can run LV EXEs (not the development environment) on CE if you buy the LV PDA module. Note that applications built using this lack features regular LV applications have.
I remember some threads dealing with the minimum necessary to run LV on linux. If I remember correctly, the basic idea is there are a few official linux versions supported and that anything outside those you will have to experiment on your own. Here's one example. You can try looking for more.
If you want, official requirements can probably be found in the main LV linux page (whereever that is).
Try to take over the world!

How to implement for sap system use HADR

hi expert ，
       i am a newbie to sap basis， we have a requirement that do HA for our sap using HADR，i want know if there are some good sulotion for my scenario。
   our scenaro is we have two window 2008 sever host，one host has a sap system and we want the sap db2 database as a primary，and the other host also has a same the system which is restore from the previous sap system which we implement by system copy using database restore not migration。i want know as our secanrio could i achive SAP application HA by HADR，if we donu2018t have HA software like MSCS。whether we must manual monitor the primary sap   when it stop because any issue like hardware failed and then manual start the other sap system in the other host？
our two sap system have different sap profile beacause the hostname are different.
our aim is when one of our host can't use we can immediate start the other sap system in the other host, the less the change the better the solution .
is it possible?
thanx very much,
best regards.

hi paul ,
    thanx for your information，i have already read the inforamtion about sg247363 once-over and SAMP。 but unfortunately we have a different situation，we only have two windows servers and must installed windows server 2008 OS because some reasons。we also don't have have other host to install sap。as this situation，how could we implement HA beacuse we also don't have shared disk。the window server are isolation。
i also read some pdf which download from sdn ， in the book the HA is implemneted as the sap application has a separate host and has two host for DB2 database using HADR，the HA is rely the cluster software 。in this situation the sap application also need HA to avoid single point failure。
    as the limited i have said above， is it possible to do HA by MSCS ，can any body tell me if the MSCS is free to install in OS windows 2008？ if we can't use it free，have any other solution？in the worst ， we must manual monitor the application and when a sap application or database can't work ，we want to restart the other sap which in the other host，we need the database synchronization between two database which using HADR。is it possible ？if it do， whether there are some additional setup for sap application because the two sap application have different sap profile name（a sap is a system copy from the other by database restore）。
    any reply will be appreciated。

Cross compile VI for embedded Linux

Hello,
I'm actually trying to port Labview to a self-designed AT91SAM9 (Atmel ARM9) platform using the "LabView for ARM" package. The way described in the porting guide is done and working for other platforms very well. But the AT91-Series seems to have some problems in combination with ULink2 Debugger and Keil uVision. It is very unstable and not deterministic.
On the other hand i've ported an Angstrom Linux to the AT91 platform wich works very well and stable. Is it possible to drop the Keil uVision Suite and use other third party compilers in combination with the LabView for ARM package?
Best regards
Kai

Hi Hai,
as far as I can see, it is neccessary to use the Keil uVision toolcain.
Take a look at the requirements of the Embedded Module for ARM microcontroller:
LabVIEW 2011 Embedded Module for ARM Microcontrollers Readme - National Instruments
http://digital.ni.com/public.nsf/allkb/FAF2376C24AB8931862578DB004C7AC3
System Requirements
The Embedded Module for ARM Microcontrollers has the following system requirements:
LabVIEW 2011 Full or Professional Development System (32-bit)
RealView Microcontroller Development Kit 4.11, including Keil µVision4
Keil ULINK2 USB-JTAG adaptor

Compiling code for LPC2378 Target using LV Embedded

Hi
I am using LabVIEW embedded for ARM 10 and using MCB2370 evaluation board. I did my evaluation successfully but now when I tried to compile the same code for an lpc2378 target (tier 2), I am getting "Error: Application is too large for target memory. Status: Build failed."
I am only doing serial communication using com 0 and using no other peripherals that are custumized for MCB2300 target.
what should I do to resolve this issue?
thanks and regards!
Mani

Hi scg,
I took a look at your file and researched a little bit about your issue, I have not been so lucky with the research since most of the doceuments are related to previous versions of the software, nevertheless it is also what makes me think that the file might be corrupted or some issue related to it so and I have found some tests that you could try:
1. Create a new FPGA file with your code on it and compile it again.
2. Turn off the firewall and compile again
3. Restart the compiler Services
Hope you might find this information helpful!
Regards,
S.Jimenez

UArch - arch linux for embeded and old machines

Hello all,
just wanted to post the new web site for uArch it can be found at
http://code.google.com/p/uarch/
check it out and let me know what you think..
zio
Last edited by zio (2008-03-31 06:21:21)

zio wrote:There back in place, if your doing an install can you use the pacman.static instructions. Just to verify the instructions are still working, should work. But I haven't tested yet
Thx for the prompt response , I've tried both methods with the static pacman and the one of my system. But unfortunately both attempts failed because of bad md5sums .
error: failed to commit transaction (corrupted package)
file gcc-4.2.1-1-i386-uclibc.pkg.tar.gz was corrupted (bad MD5 checksum)
file klibc-1.5-1-i386-uclibc.pkg.tar.gz was corrupted (bad MD5 checksum)
file klibc-udev-116-1-i386-uclibc.pkg.tar.gz was corrupted (bad MD5 checksum)
file openssl-0.9.8g-1-i386-uclibc.pkg.tar.gz was corrupted (bad MD5 checksum)
file pciutils-2.2.8-1-i386-uclibc.pkg.tar.gz was corrupted (bad MD5 checksum)
file sysfsutils-2.1.0-1-i386-uclibc.pkg.tar.gz was corrupted (bad MD5 checksum)
file vi-7.1.293-1-i386-uclibc.pkg.tar.gz was corrupted (bad MD5 checksum)

Vote for Lenovo to use Arch.

As you probably already know if you read Digg often, Lenovo is having a poll to figure which distrubtions they are going to try for adding to the market. It's being held here.(http://lenovoblogs.com/insidethebox/?p=98)
Go show your Archie some love.

Warren Watts wrote:
tardo wrote:Let them use ubuntu, and when they're tired of it, they can come to Arch =]
It's funny you should say that... I just installed Arch yesterday as a replacement for Ubuntu on one of my PC's....
Don't get me wrong, I'm still using Ubuntu on my primary PC. My family loves Ubuntu and I don't think I could convince them to change. I had a hard enough time getting my wife to make the transition from Win2k to Ubuntu!
But as tardo suggested, I grew tired of Ubuntu's somewhat unnecessary overhead on one of my older slower systems, so I decided to give Arch a try.
Me three, I've just changed 4 of my 6 family PCs over from Ubuntu to Arch. My father has a 2.6GhZ 512MB system and runs Gnome and is very happy with how much more "zippy" or responsive Arch feels.
The 2 to do are a) my server and b) my shops dial up email PC. It's an old Pentium II with only 192MB ram. It's running Gnome on Feisty really well considering. It's used by my two totally computer illiterate staff so I won't change it to Arch until I'm sure I can get the dial up over serial 1 to work. The server will change over but because it runs a lot of apps that are setup slightly differently under Arch to Ubuntu (Apache, Squirrelmail, DNS, DHCP, postfix, fetchmail, rsync, and a few others), I'll take it slowly and probably not transfer this system until I have a few days up my sleeve
Russ

Which version Diagnotics Agent to install for ERP6 systems using SM 7.1 ?

Hi,
I have a recently installed Solution Manager 7.1 system and would now like to add our managed systems for monitoring.
Our managed systems are all running NW EHP2 / ERP6 EHP5 and I have installed the latest SAPHostagent 7.2 on each of these systems. These systems are non unicode running on Suse Linux Enterprise 10.
I'm confused as to which version Diagnostic Agent to install for this setup. Under Solution Manager 7.1 in SWDC the managing agents are for a 7.3 installation only and the prerequisites require Suse Linux Enterprise 11. I'm not sure whether I can use agents located under older Solution Manager systems.
With the above information in mind can you help ?
Thanks.
Regards,
Nelis

Hi,
Please look in the table provided in below link.
http://wiki.sdn.sap.com/wiki/display/SMSETUP/Diagnostics+Agents#DiagnosticsAgents-Whichagentversionshouldbeused%3F
Regards
Vivek

Error when compiling code for crio 9068 using LV 2014 SP1

I have a project that was mostly developed in LV 2013 SP1. I recently upgraded to LV 2014 SP1, and I can no longer compile the FPGA VI. At first, I received an error stating that the Xilinx Vivado 2013.4 compilation tool was not installed. I found this tool on your site, downloaded it, and installed it on my computer. When I try to compile now, the compilation advances to a certain point, seems to hang there for a while, and then returns an error stating that toplevel_gen_propImpl.xds cannot be accessed "because it is being used by another process." I have repeated this exact same scenario on both by desktop and my laptop. I've attached a PDF of the error I windows I have received.
Attachments:
compile_error.pdf ‏114 KB

Hi scg,
I took a look at your file and researched a little bit about your issue, I have not been so lucky with the research since most of the doceuments are related to previous versions of the software, nevertheless it is also what makes me think that the file might be corrupted or some issue related to it so and I have found some tests that you could try:
1. Create a new FPGA file with your code on it and compile it again.
2. Turn off the firewall and compile again
3. Restart the compiler Services
Hope you might find this information helpful!
Regards,
S.Jimenez

Torque control is not working for my system using analog feedback

In the test bed, we are testing a brushless DC motor and a load motor generates torque to the test motor. The test motor is running with torque control scheme. They were connected together and I measured the torque signal to feedback the signal and control it. However, I just realized it only control when the torque was static. If I commanded a certain torque value (in this case, counts command), it kept generating that value. However, it did not generate the same torque in different position. I heard NI motion controller could develop torque values by mapping the torque signal with encoder signal. If it's true, the torque generation based on analog feedback, also, needs constraint by rotor position, right? W
hen I turned on the load machine with constant speed (50 RPM), the torque from the test motor did not provide the constant torque that I commanded. It generated the fluctuated torque values. This means the max torque was generated at one point and decreasing by rotating the shaft and increasing the torque. Keep going in this way. Could you please tell me if this is the correct response when we are using analog torque feedback?
Thank you.

Try the following:
1.-Set your test motor to encoder feedback instead of analog feedback.
2.-Load a velocity move on the axis for the test motor.
3.-Set the following error limit on you axis to 0 (disable following error).
4.-Use the load torque limit function to limit the amount of torque the 73xx PID loop will generate,this value is in volts, so you will need to do a correlation between the voltage output from the controller and the torque reported by your torque cell.
5.- Start the load motor.
6.-Engage the cluch.
7.-Start the velocity move on the test motor, you will see the torque ramp up to the torque limit.
8.-Call the load torque limit function on the fly to vary the torque setpoint to where you want it.

Compiling linux for embedded system using Arch.

Similar Messages

Maybe you are looking for