[etherlab-users] (kein Betreff)

Westpfahl Jasper Jasper.Westpfahl-projekt at ipk.fraunhofer.de
Tue Feb 22 15:56:13 CET 2011


Hello Etherlab community,

Has anybody had problems with memory allocation or insufficient memory running etherlab-modules made from large Simulink models?

I encounter problems running kernelmodules generated from Simulink/SimMechanics with etherlab target. After successfully compiling the model and giving the insmod command I look at the model from testmanager (connection is successful) but the model doesn't seem to run (no channel changes values ,The Value Channels/Taskinfo/0/Period/Value is 0).
I included the output of /var/log/messages after insmod and rmmod of that model into the email. Also the ethercat hardware components never enter the operational state.
When I dramatically decrease the size of the model, it works, produces reasonable output and the ethercat hardware becomes operational.

So I got a hint that for my particular model I might have to increase the rtai stack size. I did that, but any value of stack_size above roughly 120000 yields an error message saying:
insmod: error inserting 'Test2_kmod.ko': -1 Cannot allocate memory
although the command: 'free' shows 500 Mb free memory and killing numerous other processes makes no difference.

I am running matlab2009b, Etherlab 1.3 (with SimMechanics support), Ethercat master 1.5 on Suse 10.3 with a 2.6.22.5-rtai-3.6.2 Kernel

Has anybody encountered anything like this before and knows what the problem might be?

Yours faithfully,
Jasper Westpfahl
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: messages_after_insmod_small_stack.txt
URL: <http://lists.etherlab.org/pipermail/etherlab-users/attachments/20110222/ae0b3299/attachment-0006.txt>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: messages_after_rmmod_small_stack.txt
URL: <http://lists.etherlab.org/pipermail/etherlab-users/attachments/20110222/ae0b3299/attachment-0007.txt>


More information about the Etherlab-users mailing list