[Asterisk-Users] NMI issues...

Thu Jan 20 18:05:42 MST 2005

I dont' think it's the digium hardware fault.  Why?  I've got probably 8-10 
machines like this and they all get occasional NMIs with the 'uhhuh' 
message.  It's defiently kernel related...hell these boards don't work 
under anything older than 2.4.27.

--On Tuesday, December 21, 2004 00:42 -0600 "I put the Who? in Mishehu" 
<mishehu at shavedgoats.net> wrote:

> I have read thru what other users have tried in this list when they have
> experienced seemingly similar issues to what I have, without success.  I
> suspect there might be an issue regarding both the X100P and TDM04b cards
> being used in an Intel SE7525GP2 motherboard, as I had to even wait for a
> BIOS update from Intel in order to utilize my 3ware 9500 SATA raid
> controller on it (3ware Kb -->
> http://www.3ware.com/kb/article.aspx?id=12435 ).  The following is a list
> of the configuration and the issues I have encountered.  Any help would be
> greatly appreciated.
>
> System A:
> Intel SE7525GP2 motherboard.  Has BIOS revision 06, BMC 2.40 and FRUSDR
> 1.40 2x2.8GHz Xeon processors w/800Mhz FSB, actively cooled retail box set
> 1GB of ECC registered DDR333 RAM, 2 DIMMs
> 3ware 9500 SATA Raid, controlling a 3-disk raid5 array
> 550 Watt 24-pin ATX power supply
>
> System B:
> Asus A7N8X-E Deluxe
> AMD Athlon XP 3200 (barton core)
> 512 MB RAM
> ATI 9600SE OEM card
> 200GB SATA drive
>
> Process is as follows:
> 1.  On system A until further notice: The Intel SE7525GP2 conflicted with
> the firmware on the 3ware 9500 SATA RAID controller until BIOS 06 was
> released.
> 2.  After having been flashed, Slackware 10.0 was then installed on it.
> It currently runs stock kernel 2.6.9, udev 042, and the most recent
> hotplug from 2004-09-23.
> 3.  Installed a digium-supplied X101P into the machine into a 32bit PCI
> slot. 4.  Downloaded zaptel 1.0.3, compiled against kernel 2.6.9.  Upon
> modprobe wcfxo, NMI's are immediately reported (syslog to follow).  The
> instructions from README.udev were followed and parameters inserted into
> the rules file for udev.  Problem occurs regardless of whether or not
> hotplug inserts the modules or the are manually inserted by me.  Repeated
> modprobe -r wcfxo; modprobe wcfxo produces no different result as well. 5.
>  Attempted all combinations of moving the card to other compatible slots,
> CVS zaptel from 2004-12-17, and the TDM04b card.  Snippet of syslog as
> follows:
>
> Dec 19 15:33:17 nisui kernel: Freshmaker version: 71
> Dec 19 15:33:17 nisui kernel: Freshmaker passed register test
> Dec 19 15:33:18 nisui kernel: Module 0: Installed -- AUTO FXO (FCC mode)
> Dec 19 15:33:18 nisui kernel: Uhhuh. NMI received for unknown reason 31 on
> CPU 0.
> Dec 19 15:33:18 nisui kernel: Dazed and confused, but trying to continue
> Dec 19 15:33:18 nisui kernel: Do you have a strange power saving mode
> enabled?
> Dec 19 15:33:18 nisui kernel: Module 1: Installed -- AUTO FXO (FCC mode)
> Dec 19 15:33:18 nisui kernel: Module 2: Installed -- AUTO FXO (FCC mode)
> Dec 19 15:33:19 nisui kernel: Module 3: Installed -- AUTO FXO (FCC mode)
> Dec 19 15:33:19 nisui kernel: Found a Wildcard TDM: Wildcard TDM400P REV
> E/F (4 modules)
> Dec 19 15:33:19 nisui kernel: Uhhuh. NMI received for unknown reason 31 on
> CPU 0.
> Dec 19 15:33:19 nisui kernel: Dazed and confused, but trying to continue
> Dec 19 15:33:19 nisui kernel: Do you have a strange power saving mode
> enabled?
> Dec 19 15:33:19 nisui kernel: Uhhuh. NMI received for unknown reason 21 on
> CPU 0.
>
> The NMI's will continue to report reasons 21 & 31 until the module is
> removed from the kernel.
>
> 6.  Attempted several kernel reconfigurations, including ones that
> stripped out USB support and all SMP support.  Even on this machine with a
> uniprocessor kernel, NMI was still generated.
> 7.  Took the X100P and installed it in System B.  System B only reports:
> Notice: Configuration file is /etc/zaptel.conf
> line 143: Unable to open master device '/dev/zap/ctl'
> Whenever I check, /dev/zap/ctl does exist.  No NMI is generated on this
> machine, though I do not know if it is capible of doing so.  Otherwise, it
> appears to operate normally.
> 8.  Tried asterisklivecd that is maintained by somebody in Italy.
> Actually got a freshmaker error on that trial, though it appeared to be
> somewhat outdated.
>
> Just to anticipate questions, whenever the TDM04b was tested, the power
> connector was in fact connected.  An Intel Etherexpress Pro 100 ethernet
> card was tested on Machine A to verify that there was no problem with the
> slots.  The eepro100 initializes and operates perfectly.  Attached to this
> email is a copy of the kernel's configuration, and a list of currently
> loaded modules is as follows:
>
> Module                  Size  Used by
> md5                     4992  1
> ipv6                  260352  16
> hw_random               6548  0
> pciehp                 97540  0
> shpchp                101124  0
> pci_hotplug            13060  2 pciehp,shpchp
> e1000                  86788  0
> evdev                  10368  0
>
> And output from /proc/interrupts immediately after loading the wctdm
> module:           CPU0       CPU1       CPU2       CPU3
>  0:    3791068         10         10         10    IO-APIC-edge  timer 1:
>         7          1          0          0    IO-APIC-edge  i8042 9:
>     0          0          0          0   IO-APIC-level  acpi
> 12:         57          0          0          1    IO-APIC-edge  i8042 14:
>          0          1          0          0    IO-APIC-edge  ide0 169:
>   4222          0          1          0   IO-APIC-level  eth0 185:
>  0          0          0          0   IO-APIC-level  ehci_hcd 209:
> 22767          0          1          0   IO-APIC-level  3w-9xxx 217:
> 1063          0          0          1   IO-APIC-level  wctdm NMI:
> 63          0          0          0
> LOC:    3791179    3791182    3791180    3791179
> ERR:          0
> MIS:          0
>
> Thanks in advance,
>
> mishehu

--
GPG/PGP --> 0xE736BD7E 5144 6A2D 977A 6651 DFBE 1462 E351 88B9 E736 BD7E 