Error executing SSH command doing a embedded code on the BeagleBoard Rev. C from Matlab 2010a in a Windows host

GoogleGroups
1

Greetings!!

We are working with the BeagleBoard revision C3, running the Angstrom Linux distribution
root@beagleboard:~# uname -a
Linux beagleboard 2.6.32 #3 PREEMPT Tue Jul 26 10:56:56 CEST 2011 armv7l unknown

We want to do:
A embedded coder from a Matlab model, to generate code on a Windows XP host, and then build it remotely on a BeagleBoard running Linux.

We have istalled:
Following the Matlab help, we have installed Java, Eclipse and the GNU Tool Chain on a windows host. In the same way we installed Plink (a command-line interface to the PuTTY back ends) and PSCP (an SCP client, i.e. command-line secure file copy) just a PuTTY utilities to do a comunication.
We have verifyed the GNU Tool Chain on Angstrom in the BeagleBoard.
And following the instruction from Matlab we have installed this software:
root@beagleboard:~# opkg install binutils
root@beagleboard:~# opkg install gcc
root@beagleboard:~# opkg install gcc-symlinks
root@beagleboard:~# opkg install cpp-symlinks
root@beagleboard:~# opkg install libstdc+±dev
root@beagleboard:~# opkg install make-dev

We did it:
Continuing with the project, I have made a simple model in Simulink, which I have attached (call_stats_block2.mdl), with the Target Preferences Block, and following the Matlab example I configured it to generate the code. The following lines are from the help of Matlab 2010a

“Help: Embedded code/User’s Guide/Embedded IDEs and Embedded Targets/Working with Linux Target/Example: Build Generated Code on a BeagleBoard Running Linux”

  1. Enter xmakefilesetup at the MATLAB command line. In the XMakefile Configuration dialog, set template to gmake and configuration to gcc_target.
  2. For the Target Preferences block in your model, set IDE/Toolchain to Eclipse, and set the Board to BeagleBoard ARM.
  3. In your model, open Simulation > Configuration Settings. Under Code Generation, select the IDE Link pane. Set Build format to Makefile, and set Build Action to Create_makefile.
    4.Generate the code by pressing Ctrl B.
    5.Use the load method to load the buildInfo.mat file from the project directory to the BeagleBoard. For example, at the MATLAB command prompt enter:

s =load(‘sumdiff_codegen_eclipseide\buildInfo.mat’)

  1. Use remoteBuild function to build the code on the BeagleBoard. For example, enter:

remoteBuild(s.buildInfo, ‘/home/root’, ‘144.212.110.193’, ‘root’, ‘password’, ‘C:\utils\putty’)
The GNU compiler toolchain automatically builds and runs the software on the BeagleBoard.

The problem:
The problem which I have is in the 6 step: “Error executing SSH command:The system can not find the path specified.”

Thanks in advance for any tips!!

2

Greetings!!

We are working with the BeagleBoard revision C3, running the Angstrom Linux
distribution
root@beagleboard:~# uname -a
Linux beagleboard 2.6.32 #3 PREEMPT Tue Jul 26 10:56:56 CEST 2011 armv7l
unknown

We want to do:
A embedded coder from a Matlab model, to generate code on a Windows XP host,
and then build it remotely on a BeagleBoard running Linux.

We have istalled:
Following the Matlab help, we have installed Java, Eclipse and the GNU Tool
Chain on a windows host. In the same way we installed Plink (a command-line
interface to the PuTTY back ends) and PSCP (an SCP client, i.e. command-line
secure file copy) just a PuTTY utilities to do a comunication.
We have verifyed the GNU Tool Chain on Angstrom in the BeagleBoard.
And following the instruction from Matlab we have installed this software:
root@beagleboard:~# opkg install binutils
root@beagleboard:~# opkg install gcc
root@beagleboard:~# opkg install gcc-symlinks
root@beagleboard:~# opkg install cpp-symlinks
root@beagleboard:~# opkg install libstdc++-dev
root@beagleboard:~# opkg install make-dev

We did it:
Continuing with the project, I have made a simple model in Simulink, which I
have attached (call_stats_block2.mdl), with the Target Preferences Block,
and following the Matlab example I configured it to generate the code. The
following lines are from the help of Matlab 2010a

“Help: Embedded code/User’s Guide/Embedded IDEs and Embedded Targets/Working
with Linux Target/Example: Build Generated Code on a BeagleBoard Running
Linux”

1. Enter xmakefilesetup at the MATLAB command line. In the XMakefile
Configuration dialog, set template to gmake and configuration to gcc_target.
2. For the Target Preferences block in your model, set IDE/Toolchain to
Eclipse, and set the Board to BeagleBoard ARM.
3. In your model, open Simulation > Configuration Settings. Under Code
Generation, select the IDE Link pane. Set Build format to Makefile, and set
Build Action to Create_makefile.
4.Generate the code by pressing Ctrl B.
5.Use the load method to load the buildInfo.mat file from the project
directory to the BeagleBoard. For example, at the MATLAB command prompt
enter:
>> s =load(‘sumdiff_codegen_eclipseide\buildInfo.mat')
6. Use remoteBuild function to build the code on the BeagleBoard. For
example, enter:
>> remoteBuild(s.buildInfo, '/home/root', '144.212.110.193', 'root',
'password', 'C:\utils\putty')
The GNU compiler toolchain automatically builds and runs the software on the
BeagleBoard.

The problem:
The problem which I have is in the 6 step: "Error executing SSH command:The
system can not find the path specified."

Thanks for the fairly well written problem statement and issue description.

Have you tried simply logging in with putty to make sure that part
works? Putty is an terminal package that uses an ssh connection.
You'll need to make sure each of the arguments you pass into
remoteBuild() are valid.

I don't know anything about s.buildInfo, but I'd guess that step 5
sets up 's' properly.

I'm guessing that '/home/root' is the target directory in which you
expect to run. That's probably fine.

I'm guessing the next parameter is the BeagleBoard's IP address. You
should verify that your BeagleBoard is on the network and actually has
that IP address.

'root' is probably the login username. That'll be fine.

'password' is probably the login password. For the default Angstrom
builds, that'll be wrong. Switch it to a blank password, ie. ''.

'C:\utils\putty' is where to find Putty that would be used to invoke
an ssh connection. Make sure that the executable is actually in that
location on your computer. Try executing it stand-alone using the IP
address, username and password. You should get to a Linux command
prompt.

Hope that helps!

3

hello
thank you very much, this works perfect, but now we have this problem
"cc: no input files"
I dont know whats going on, I have done: open Simulation >
Configuration Settings. Under Code Generation, select the IDE Link
pane. Set Build format to Makefile, and set Build Action to
Create_makefile. On my model
thanks a lot for the help

The complete command is:

remoteBuild(s.buildInfo, '/home/root', '192.168.60.243', 'root',
'root', 'C:\utils')
### Adding sources to archive
### Adding include files to archive
### Adding libraries to archive
### Creating remote project directory
### Creating file archive
### Copying archive to remote target
### Extracting archive on remote target
### Building executable on remote target
make: Entering directory `/home/root/call_stats_block2_eclipseide'
cc -o .o
make: Leaving directory `/home/root/call_stats_block2_eclipseide'
make: *** [.o] Error 1
??? Error using ==> remoteBuild>executeRemoteCommand at 198
Error executing SSH command: make: Entering directory `/home/root/
call_stats_block2_eclipseide'
cc -o .o
make: Leaving directory `/home/root/call_stats_block2_eclipseide'
make: *** [.o] Error 1

Error in ==> remoteBuild at 123
executeRemoteCommand(cmd, true);

4

Hello All,
I am facing the same problem. Is anyone how to solve this issue? plz guide…

Thanks & Regards,
Avadhut
" The Will of God will never take you where the Grace of God will not protect you."