[etherlab-users] Kernel question

[Kaashoek, I. (Irene)]

Hi,

First of all, thank you Erwin for the files.
I continued working on this topic but I had some problems:

- Since the newest RTAI patch is 2.6.32.2, I used this kernel. This did not give any problems with the 8139too driver, but it did using the e1000e driver. I took the master of Erwin, and the ethercat-patched driver from the newest master version (of kernel 2.6.31) and changed it as good as I could to something that would be compatible with both the master and the driver. The result did compile, but as I used it, it would give the same error as with the master I used before, and as Newell Jensen mentioned in his email [2.6.31 RT_PREEMPT kernel and e1000 driver]: the ethercat master seems to run, but keeps waiting for the device. (so in this case, I used the ec_e1000e.ko module as suggested in the email from Erwin Burgstaller)

- As Andreas suggested to Newell, I then tried to use the generic driver, allthough the manual says this is not possible with RTAI. This time the ethercat master runs and accepts the generic driver, but when I insert a module, things go wrong. The slaves do not go to operational state as expected and I get the following errors (from dmesg):
Successfully registered application "test1" with RT-AppCore.
[ 4192.316550] ------------[ cut here ]------------
[ 4192.316556] WARNING: at kernel/sched.c:5316 add_preempt_count+0x7d/0xb0()
[ 4192.316558] Hardware name: OEM
[ 4192.316560] Modules linked in: test1_kmod rt_appcore rtai_math rtai_sem rtai_sched rtai_hal ec_generic ec_master nls_iso8859_1 nls_cp437 vfat fat usb_storage binfmt_misc ppdev fbcon tileblit font bitblit softcursor i915 drm_kms_helper lp intel_agp drm e1000e i2c_algo_bit agpgart parport usbhid hid serio_raw [last unloaded: ec_master]
[ 4192.316594] Pid: 0, comm: swapper Not tainted 2.6.32.2-rtai #1
[ 4192.316596] Call Trace:
[ 4192.316599]  [<c0144432>] warn_slowpath_common+0x72/0xa0
[ 4192.316604] Default Trap Handler: vector 14: Suspend RT task f2c65140
[ 4192.316610] ------------[ cut here ]------------
[ 4192.316612] WARNING: at kernel/softirq.c:143 local_bh_enable+0x88/0xb0()
[ 4192.316614] Hardware name: OEM
[ 4192.316616] Modules linked in: test1_kmod rt_appcore rtai_math rtai_sem rtai_sched rtai_hal ec_generic ec_master nls_iso8859_1 nls_cp437 vfat fat usb_storage binfmt_misc ppdev fbcon tileblit font bitblit softcursor i915 drm_kms_helper lp intel_agp drm e1000e i2c_algo_bit agpgart parport usbhid hid serio_raw [last unloaded: ec_master]
[ 4192.316645] Pid: 0, comm: swapper Not tainted 2.6.32.2-rtai #1
[ 4192.316647] Call Trace:
[ 4192.316650]  [<c0144432>] warn_slowpath_common+0x72/0xa0

I do not really understand what is happening here, but apparently something is going wrong...
To do this, I used another master release as the one from Erwin Burgstaller, since the newest version of etherlab did not work with this master (not the newest master, since this one already gave me a lot of trouble on another computer).

My suggestion to solve the problems would be to compile another preempt kernel, kernel 2.6.29  and then use the master and driver files from Erwin Burgstaller. Then I would perhaps need to find a slightly older version of etherlab so that it is compatible with the master.

Does anyone have any better suggestions or advice to make my system work? Any help would be appreciated very much.

Best regards,
Irene Kaashoek

Btw I am using a Beckhoff industrial pc with PCIE network cards: 82574L and 82567LM.

-----Original Message-----
From: etherlab-users-bounces at etherlab.org [mailto:etherlab-users-bounces at etherlab.org] On Behalf Of Erwin Burgstaller
Sent: vrijdag 15 april 2011 12:57
To: etherlab-users at etherlab.org
Subject: Re: [etherlab-users] Kernel question


Hi!

* Fri, Apr 15, 2011 - 09:03:41 +0000, Kaashoek I. Irene - irene.kaashoek at tno.nl:
> Hello,
> 
> Currently I am trying to set up Etherlab with kernel 2.6.32.2 and 
> master 1.5 with the e1000e driver. I would like to take a look at the 
> files Mr Burgstaller posted, but the links seems to be broken.
> Could anybody help me with a new link to these files or any other help?
> Else I will switch to the generic driver.

This is the complete master as we currently use it with kernel 2.6.29.5, stored on a much better and more durable place now:

http://dl.dropbox.com/u/26324125/ecmaster-v1_5-2.6.29.5-xenomai.tar.bz2

There's another type of e1000 out there, which isn't yet recognized nor tested. If you use these card, you will have to extend the old driver with the signature of that type.

Best regards,
  Erwin

PS: I've recognized some mails beeing swallowed by spamgourmet, probably have been sent to me directly. This spamgourmet address accepts mails to the lists only.

If you like to create an account on dropbox.com, please use this URL: http://db.tt/w1WnJqb ;-)

--
Erwin Burgstaller
_______________________________________________
etherlab-users mailing list
etherlab-users at etherlab.org
http://lists.etherlab.org/mailman/listinfo/etherlab-users
This e-mail and its contents are subject to the DISCLAIMER at http://www.tno.nl/emaildisclaimer