* Re: [Xenomai] process behaviour when running xenomai
2014-03-06 9:46 [Xenomai] process behaviour when running xenomai Huub Van Niekerk
@ 2014-03-06 16:27 ` Gilles Chanteperdrix
2014-03-06 19:21 ` Gilles Chanteperdrix
` (2 subsequent siblings)
3 siblings, 0 replies; 13+ messages in thread
From: Gilles Chanteperdrix @ 2014-03-06 16:27 UTC (permalink / raw)
To: Huub Van Niekerk; +Cc: xenomai@xenomai.org
On 03/06/2014 10:46 AM, Huub Van Niekerk wrote:
> Hi,
>
> I'm running Xenomai 2.6.3 with kernel 3.10.18 now on BeagleBoard-Xm
> Rev. C and experience this behaviour: when I try to install packages
> from commandline, the installation manager starts quick and reads the
> package(s), but after that it stays dead quiet like it's frozen. ps
> -ef shows it's just going on, no defunc or anything like that. Also,
> when I want to watch the processes using top, it starts, but nothing
> shows. When I want to end it, Ctl-C gives no response giving me no
> other option than to use kill. Is this a known behaviour when running
> Xenomai ?
No, obviously not. What does "dmesg" say when you are in that state?
--
Gilles.
^ permalink raw reply [flat|nested] 13+ messages in thread* Re: [Xenomai] process behaviour when running xenomai
2014-03-06 9:46 [Xenomai] process behaviour when running xenomai Huub Van Niekerk
2014-03-06 16:27 ` Gilles Chanteperdrix
@ 2014-03-06 19:21 ` Gilles Chanteperdrix
2014-03-07 10:14 ` Robert Berger
2014-03-07 16:31 ` Gilles Chanteperdrix
3 siblings, 0 replies; 13+ messages in thread
From: Gilles Chanteperdrix @ 2014-03-06 19:21 UTC (permalink / raw)
To: Huub Van Niekerk; +Cc: xenomai@xenomai.org
On 03/06/2014 10:46 AM, Huub Van Niekerk wrote:
> Hi,
>
> I'm running Xenomai 2.6.3 with kernel 3.10.18 now on BeagleBoard-Xm
> Rev. C and experience this behaviour: when I try to install packages
> from commandline, the installation manager starts quick and reads the
> package(s), but after that it stays dead quiet like it's frozen. ps
> -ef shows it's just going on, no defunc or anything like that. Also,
> when I want to watch the processes using top, it starts, but nothing
> shows. When I want to end it, Ctl-C gives no response giving me no
> other option than to use kill. Is this a known behaviour when running
> Xenomai ?
It seems that you have a lot of things enabled in your kernel
configuration, could you post the kernel configuration somewhere?
Have you tried the exact same configuration, only without xenomai?
Regards.
--
Gilles.
^ permalink raw reply [flat|nested] 13+ messages in thread* Re: [Xenomai] process behaviour when running xenomai
2014-03-06 9:46 [Xenomai] process behaviour when running xenomai Huub Van Niekerk
2014-03-06 16:27 ` Gilles Chanteperdrix
2014-03-06 19:21 ` Gilles Chanteperdrix
@ 2014-03-07 10:14 ` Robert Berger
2014-03-07 12:57 ` Gilles Chanteperdrix
2014-03-07 13:21 ` Gilles Chanteperdrix
2014-03-07 16:31 ` Gilles Chanteperdrix
3 siblings, 2 replies; 13+ messages in thread
From: Robert Berger @ 2014-03-07 10:14 UTC (permalink / raw)
To: xenomai
Hi,
On 03/06/2014 11:46 AM, Huub Van Niekerk wrote:
> Hi,
>
> I'm running Xenomai 2.6.3 with kernel 3.10.18 now on BeagleBoard-Xm Rev. C and experience this behaviour: when I try to install packages from commandline, the installation manager starts quick and reads the package(s), but after that it stays dead quiet like it's frozen. ps -ef shows it's just going on, no defunc or anything like that. Also, when I want to watch the processes using top, it starts, but nothing shows. When I want to end it, Ctl-C gives no response giving me no other option than to use kill. Is this a known behaviour when running Xenomai ?
I tried something similar (just took the patch from the ELDK 5.5) but
have some issues:
1) no network on the beagle-xm and even if I hack my way through with
phy_nop it's not getting better
I really would like to boot with a rootfs over nfs and it does not look
like this is going to hapen with a 3.10 kernel on a beagle-xm.
3.13.x with fdt works nicely, but there is no Xenomai patch for it;)
>
>
> Thanks,
>
> hvn
>
Regards,
Robert
...For every complex problem there is a solution which is simple, neat
and wrong. -- H.L. Mencken
My public pgp key is available,at:
http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x90320BF1
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [Xenomai] process behaviour when running xenomai
2014-03-07 10:14 ` Robert Berger
@ 2014-03-07 12:57 ` Gilles Chanteperdrix
2014-03-07 13:21 ` Gilles Chanteperdrix
1 sibling, 0 replies; 13+ messages in thread
From: Gilles Chanteperdrix @ 2014-03-07 12:57 UTC (permalink / raw)
To: Robert Berger; +Cc: xenomai
On 03/07/2014 11:14 AM, Robert Berger wrote:
> Hi,
>
> On 03/06/2014 11:46 AM, Huub Van Niekerk wrote:
>> Hi,
>>
>> I'm running Xenomai 2.6.3 with kernel 3.10.18 now on BeagleBoard-Xm Rev. C and experience this behaviour: when I try to install packages from commandline, the installation manager starts quick and reads the package(s), but after that it stays dead quiet like it's frozen. ps -ef shows it's just going on, no defunc or anything like that. Also, when I want to watch the processes using top, it starts, but nothing shows. When I want to end it, Ctl-C gives no response giving me no other option than to use kill. Is this a known behaviour when running Xenomai ?
>
> I tried something similar (just took the patch from the ELDK 5.5) but
> have some issues:
>
> 1) no network on the beagle-xm and even if I hack my way through with
> phy_nop it's not getting better
Have you selected an USB transceiver? This is a difference between 3.8
and 3.10, and is not enabled by default.
--
Gilles.
^ permalink raw reply [flat|nested] 13+ messages in thread* Re: [Xenomai] process behaviour when running xenomai
2014-03-07 10:14 ` Robert Berger
2014-03-07 12:57 ` Gilles Chanteperdrix
@ 2014-03-07 13:21 ` Gilles Chanteperdrix
2014-03-07 21:14 ` Robert Berger
1 sibling, 1 reply; 13+ messages in thread
From: Gilles Chanteperdrix @ 2014-03-07 13:21 UTC (permalink / raw)
To: Robert Berger; +Cc: xenomai
On 03/07/2014 11:14 AM, Robert Berger wrote:
> Hi,
>
> On 03/06/2014 11:46 AM, Huub Van Niekerk wrote:
>> Hi,
>>
>> I'm running Xenomai 2.6.3 with kernel 3.10.18 now on BeagleBoard-Xm Rev. C and experience this behaviour: when I try to install packages from commandline, the installation manager starts quick and reads the package(s), but after that it stays dead quiet like it's frozen. ps -ef shows it's just going on, no defunc or anything like that. Also, when I want to watch the processes using top, it starts, but nothing shows. When I want to end it, Ctl-C gives no response giving me no other option than to use kill. Is this a known behaviour when running Xenomai ?
>
> I tried something similar (just took the patch from the ELDK 5.5) but
> have some issues:
>
> 1) no network on the beagle-xm and even if I hack my way through with
> phy_nop it's not getting better
>
> I really would like to boot with a rootfs over nfs and it does not look
> like this is going to hapen with a 3.10 kernel on a beagle-xm.
FWIW, I only test xenomai with tftp + nfs, even the U-boot script is not
on the board itself. Xenomai 2.6.3 with linux 3.10.18 has been tested on
the pandaboard, which has exactly the same ethernet chip as beagle-xm
(from what I see in Huub's dmesg). The only difficulty for me was the
missing USB transceiver in the kernel configuration.
Generally, before you report a non-working configuration of Xenomai,
with a certain version of the kernel, please try the exact same kernel
configuration without Xenomai.
--
Gilles.
^ permalink raw reply [flat|nested] 13+ messages in thread* Re: [Xenomai] process behaviour when running xenomai
2014-03-07 13:21 ` Gilles Chanteperdrix
@ 2014-03-07 21:14 ` Robert Berger
2014-03-07 21:49 ` Gilles Chanteperdrix
2014-03-08 11:14 ` Huub Van Niekerk
0 siblings, 2 replies; 13+ messages in thread
From: Robert Berger @ 2014-03-07 21:14 UTC (permalink / raw)
To: xenomai
Hi Gilles,
On 03/07/2014 03:21 PM, Gilles Chanteperdrix wrote:>
> Generally, before you report a non-working configuration of Xenomai,
> with a certain version of the kernel, please try the exact same kernel
> configuration without Xenomai.
>
>
I think I was not explicit enough in my previous email.
The problem that ethernet does not work is __NOT__ due to Xenomai!
It's a general problem with beagle-xm and 3.10 (and not just)
The workaround in theory is CONFIG_NOP_USB_XCEIV=M which will in best
case turn on the ethernet interface in user space and not allow the
kernel to load a rootfs over nfs.
My intention was to ask the op if he has the network up and running with
a 3.10 kernel and Xenomai.
Sorry for not being clear enough.
Regards,
Robert
...There's no time to stop for gas, we're already late.- Karin Donker
My public pgp key is available,at:
http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x90320BF1
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [Xenomai] process behaviour when running xenomai
2014-03-07 21:14 ` Robert Berger
@ 2014-03-07 21:49 ` Gilles Chanteperdrix
2014-03-08 11:14 ` Huub Van Niekerk
1 sibling, 0 replies; 13+ messages in thread
From: Gilles Chanteperdrix @ 2014-03-07 21:49 UTC (permalink / raw)
To: Robert Berger; +Cc: xenomai
On 03/07/2014 10:14 PM, Robert Berger wrote:
> Hi Gilles,
>
> On 03/07/2014 03:21 PM, Gilles Chanteperdrix wrote:>
>> Generally, before you report a non-working configuration of Xenomai,
>> with a certain version of the kernel, please try the exact same kernel
>> configuration without Xenomai.
>>
>>
>
> I think I was not explicit enough in my previous email.
> The problem that ethernet does not work is __NOT__ due to Xenomai!
>
> It's a general problem with beagle-xm and 3.10 (and not just)
> The workaround in theory is CONFIG_NOP_USB_XCEIV=M which will in best
> case turn on the ethernet interface in user space and not allow the
> kernel to load a rootfs over nfs.
I would say you need CONFIG_NOP_USB_XCEIV=Y instead of M, otherwise you
will need to put the module in an initramfs (if you try to load the
module from the rootfs on NFS, you have an obvious chicken-and-egg issue).
I also turn on the OMAP-specific transceivers in the configuration I
use, just in case.
You may also have to use the rootwait kernel parameter in order to wait
for the USB detection.
And finally, in u-boot, you may want to use "usb stop" before you call
bootm, as some versions of Linux do not like to start with the USB
already running.
--
Gilles.
^ permalink raw reply [flat|nested] 13+ messages in thread* Re: [Xenomai] process behaviour when running xenomai
2014-03-07 21:14 ` Robert Berger
2014-03-07 21:49 ` Gilles Chanteperdrix
@ 2014-03-08 11:14 ` Huub Van Niekerk
2014-03-22 16:03 ` Robert Berger
1 sibling, 1 reply; 13+ messages in thread
From: Huub Van Niekerk @ 2014-03-08 11:14 UTC (permalink / raw)
To: Robert Berger, xenomai@xenomai.org
On Friday, March 7, 2014 10:57 PM, Robert Berger <gmane@reliableembeddedsystems.com> wrote:
Hi Gilles,
On 03/07/2014 03:21 PM, Gilles Chanteperdrix wrote:>
> Generally, before you report a non-working configuration of Xenomai,
> with a certain version of the kernel, please try the exact same kernel
> configuration without Xenomai.
>
>
>I think I was not explicit enough in my previous email.
>The problem that ethernet does not work is __NOT__ due to Xenomai!
>It's a general problem with beagle-xm and 3.10 (and not just)
>The workaround in theory is CONFIG_NOP_USB_XCEIV=M which will in best
>case turn on the ethernet interface in user space and not allow the
>kernel to load a rootfs over nfs.
>My intention was to ask the op if he has the network up and running with
>a 3.10 kernel and Xenomai.
Hi Robert,
Both with and without Xenomai, my network runs fine. For me, the breakpoint of getting it to work was the enabling ([*] in kernel config) of the 95XX chipset, since this chipset combines USB-host and ethernet-over-USB. So it is not real ethernet but simulated. See the xM-manual and 9514 data, which is not in the manual. I have set the NOP_USB_XCEIV = [y] btw. For completeness, I run kernel 3.10.18, since that is by my knowledge the currently latest supported kernel by Xenomai.
Regards,
hvn
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [Xenomai] process behaviour when running xenomai
2014-03-08 11:14 ` Huub Van Niekerk
@ 2014-03-22 16:03 ` Robert Berger
0 siblings, 0 replies; 13+ messages in thread
From: Robert Berger @ 2014-03-22 16:03 UTC (permalink / raw)
To: xenomai
Hi,
On 03/08/2014 12:14 PM, Huub Van Niekerk wrote:>
> Hi Robert,
>
> Both with and without Xenomai, my network runs fine. For me, the
breakpoint of getting it to work was the enabling ([*] in kernel config)
of the 95XX chipset, since this chipset combines USB-host and
ethernet-over-USB. So it is not real ethernet but simulated. See the
xM-manual and 9514 data, which is not in the manual. I have set the
NOP_USB_XCEIV = [y] btw. For completeness, I run kernel 3.10.18, since
that is by my knowledge the currently latest supported kernel by Xenomai.
Can you please post (e.g. pastebin) your kernel config somewhere? I will
be busy next week, but definitely want to give it a try.
If you want to try 3.10.28 the patch is here [1]
Regards,
Robert
[1]
http://git.denx.de/?p=eldk.git;a=blob;f=meta-eldk/recipes-kernel/xenomai/files/0001-ipipe-kernel-3.10.patch;h=a8be9819c1df055db348c5d2f7278cbd43cb8d6c;hb=eldk-rel-v5.5
>
> Regards,
>
> hvn
>
...To go faster, slow down. Everybody who knows about orbital mechanics
understands that.- Scott Cherf
My public pgp key is available,at:
http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x90320BF1
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [Xenomai] process behaviour when running xenomai
2014-03-06 9:46 [Xenomai] process behaviour when running xenomai Huub Van Niekerk
` (2 preceding siblings ...)
2014-03-07 10:14 ` Robert Berger
@ 2014-03-07 16:31 ` Gilles Chanteperdrix
2014-03-07 19:48 ` Huub Van Niekerk
3 siblings, 1 reply; 13+ messages in thread
From: Gilles Chanteperdrix @ 2014-03-07 16:31 UTC (permalink / raw)
To: Huub Van Niekerk; +Cc: xenomai@xenomai.org
On 03/06/2014 10:46 AM, Huub Van Niekerk wrote:
> Hi,
>
> I'm running Xenomai 2.6.3 with kernel 3.10.18 now on BeagleBoard-Xm
> Rev. C and experience this behaviour: when I try to install packages
> from commandline, the installation manager starts quick and reads the
> package(s), but after that it stays dead quiet like it's frozen. ps
> -ef shows it's just going on, no defunc or anything like that. Also,
> when I want to watch the processes using top, it starts, but nothing
> shows. When I want to end it, Ctl-C gives no response giving me no
> other option than to use kill. Is this a known behaviour when running
> Xenomai ?
Does the access to SD-card work fine? BeagleBone users have reported an
issue with the OMAP SD card controller fixed in the 3.12 kernel by this
patch:
http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=7472bab236bdee1173412585591329e718f4d324
Patching the kernel with Xenomai made the bug more likely to happen.
--
Gilles.
^ permalink raw reply [flat|nested] 13+ messages in thread* Re: [Xenomai] process behaviour when running xenomai
2014-03-07 16:31 ` Gilles Chanteperdrix
@ 2014-03-07 19:48 ` Huub Van Niekerk
2014-03-07 19:53 ` Gilles Chanteperdrix
0 siblings, 1 reply; 13+ messages in thread
From: Huub Van Niekerk @ 2014-03-07 19:48 UTC (permalink / raw)
To: Gilles Chanteperdrix; +Cc: xenomai@xenomai.org
On Friday, March 7, 2014 5:33 PM, Gilles Chanteperdrix <gilles.chanteperdrix@xenomai.org> wrote:
On 03/06/2014 10:46 AM, Huub Van Niekerk wrote:
> Hi,
>
> I'm running Xenomai 2.6.3 with kernel 3.10.18 now on BeagleBoard-Xm
> Rev. C and experience this behaviour: when I try to install packages
> from commandline, the installation manager starts quick and reads the
> package(s), but after that it stays dead quiet like it's frozen. ps
> -ef shows it's just going on, no defunc or anything like that. Also,
> when I want to watch the processes using top, it starts, but nothing
> shows. When I want to end it, Ctl-C gives no response giving me no
> other option than to use kill. Is this a known behaviour when running
> Xenomai ?
>Does the access to SD-card work fine? BeagleBone users have reported an
>issue with the OMAP SD card controller fixed in the 3.12 kernel by this
>patch:
>http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=7472bab236bdee1173412585591329e718f4d324
>Patching the kernel with Xenomai made the bug more likely to happen.
Don't know if it has anything to do with it, but quite a few write actions seem to end in stalling. But so do other actions, like sudo. Via another ssh session, I can then see what's happening and either kill the action or let it go on. Quite annoying and causes a lot of delay in waiting and then see if something is still getting on.
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [Xenomai] process behaviour when running xenomai
2014-03-07 19:48 ` Huub Van Niekerk
@ 2014-03-07 19:53 ` Gilles Chanteperdrix
0 siblings, 0 replies; 13+ messages in thread
From: Gilles Chanteperdrix @ 2014-03-07 19:53 UTC (permalink / raw)
To: Huub Van Niekerk; +Cc: xenomai@xenomai.org
On 03/07/2014 08:48 PM, Huub Van Niekerk wrote:
> On Friday, March 7, 2014 5:33 PM, Gilles Chanteperdrix
> <gilles.chanteperdrix@xenomai.org> wrote:
>
> On 03/06/2014 10:46 AM, Huub Van Niekerk wrote:
>
>> Hi,
>>
>> I'm running Xenomai 2.6.3 with kernel 3.10.18 now on
>> BeagleBoard-Xm Rev. C and experience this behaviour: when I try to
>> install packages from commandline, the installation manager starts
>> quick and reads the package(s), but after that it stays dead quiet
>> like it's frozen. ps -ef shows it's just going on, no defunc or
>> anything like that. Also, when I want to watch the processes using
>> top, it starts, but nothing shows. When I want to end it, Ctl-C
>> gives no response giving me no other option than to use kill. Is
>> this a known behaviour when running Xenomai ?
>
>> Does the access to SD-card work fine? BeagleBone users have
>> reported an issue with the OMAP SD card controller fixed in the
>> 3.12 kernel by this patch:
>
>> http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=7472bab236bdee1173412585591329e718f4d324
>
>> Patching the kernel with Xenomai made the bug more likely to
>> happen.
>
> Don't know if it has anything to do with it, but quite a few write
> actions seem to end in stalling.
I guess this is what you would observe with a rootfs on SD card if the
SD card driver was at fault. In any case, this is not normal, and this
is what you should investigate.
--
Gilles.
^ permalink raw reply [flat|nested] 13+ messages in thread