Re: [Xenomai-help] Xenomai LTTng support

[Jan Kiszka <jan.kiszka@domain.hid>]

Vikesh Rambaran wrote:
> On Tue, 2009-04-14 at 13:53 +0200, Jan Kiszka wrote:
>> Vikesh Rambaran wrote:
>>> On Tue, 2009-04-14 at 13:35 +0200, Jan Kiszka wrote:
>>>>>> NMI watchdog, see Xenomai machine kconfig menu, use nmi_watchdog=2 to
>>>>>> finally enable it at the kernel command line.
>>>>>>
>>>>> There are build issues with NMI enabled.
>>>>>
>>>>> Would you prefer testing with the latest git updates, or can i just 
>>>>> replace the following in hal32.c (line 139)
>>>>>
>>>>> - die_nmi(regs, buf); 
>>>>> + die_nmi(buf, regs, 1);
>>>> If your CPU is not too new, this may suffice. Otherwise you should
>>>> switch over to the development head (upcoming 2.5) as there we have
>>>> up-to-date NMI support.
>>>> BTW, trying 2.5-devel might be useful anyway to check if the picture
>>>> changes (maybe we even see a clearer error with that branch...).
>>>>
>>> Ok, thanx. I have a Core Duo and a Core 2. Will try the 'short' approach
>>> first as our external git access at work is blocked. 
> 
> Tried the test with NMI enabled and die_nmi(buf, regs, 1);
> No additional Xenomai / IPIPE info is displayed on the serial console.
> 
> What type of response were we hoping for ? Reset ?

Nope. Some error message that the NMI watchdog triggered. You should see
NMI increases in /proc/interrupts and /proc/xenomai/irq. But if you are
on Intel Core processors, you need the patches in git head. 2.4.7 does
not support your CPUs.

> 
> There is occasionally the following message that appears on the serial
> console after logging into the gnome desktop (not sure if it's relevant
> to this issue) :
> 
> evdev.c(EVIOCGBIT): Suspicious buffer size 511, limiting output to 64
> bytes. See http://userweb.kernel.org/~dtor/eviocgbit-bug.html
> 

Doesn't look directly related.

Jan

-- 
Siemens AG, Corporate Technology, CT SE 2
Corporate Competence Center Embedded Linux