[OpenIndiana-discuss] PCIEX-8000-KP was Re: Nvidia errors on 151a4

[Bryan N Iotti]

I was wrong on that, they went away for a couple of hours and then we 
were back to square one.

I followed instructions on 
http://www.mail-archive.com/fm-discuss@opensolaris.org/msg00298.html
and added

pcie_ce_mask=-1;



to both the nvidia.conf file and the pcieb.conf in /kernel/drv.

The system has been stable for several hours now, I have monitored the 
fmdump -e output often and there have been no more occurrences.

Bryan

On 05/ 9/12 12:41 PM, Bryan N Iotti wrote:
> To answer my own question, I have finally solved the issue proper (I 
> hope).
>
> Turns out that in the BIOS there was a feature I knew nothing about, 
> and HP never mentions.
>
> Turn off PCI SERR# Generation and all the fabric issues vanish.
>
> Hope this helps someone avoid the HELL it was to try and do anything 
> while the computer was crashing hard all the time.
>
> Bryan
>
> On 05/ 9/12 11:13 AM, Bryan Iotti wrote:
>> Hi all,
>>
>> The reinstall went smoothly, but upon reinstall, fmd started to complain
>> about problems with the only PCI Express x16 slot (slot 2) that holds 
>> the
>> video card.
>>
>> It is littering the error log with reports of problems with the pci 
>> fabric
>> and it freezes up shortly thereafter. Before the reinstall, I only 
>> had the
>> errors in fmdump -e, the one in fmadm faulty had gone away after 
>> removing
>> and cleaning the old dust-filled Quadro.
>>
>> This is a new card in the slot, an Asus Nvidia GTX 550 Ti.
>>
>> I have read up on the error:
>> http://www.illumos.org/msg/PCIEX-8000KP
>>
>> On the Oracle blogs, it says that sometimes it's a fake, reported by fmd
>> because of a bug they suppositely fixed in Sol 10 11/08 (I think).
>>
>> Now, my course of action was to try to disable the fmd modules 
>> relating to
>> fabric and io by issuing fmadm disable io-report and fmdadm disable
>> fabric-xlate, but to no avail, it still crashed and they are reloaded on
>> reboot. I then tried to disable the fmd service in SMF. Crashed still.
>>
>> Any clues? I know the card is mounted properly and has enough power 
>> through
>> the additional cable that runs from the PSU to the accessory 6 pin plug.
>>
>> Could it be that it's my motherboard that's kicking the can this time?
>>
>> Bryan
>> On May 9, 2012 9:32 AM, "Bryan Iotti"<ironsides.medvet at gmail.com>  
>> wrote:
>>
>>> After a lot of time spent troubleshooting, only to see the crashes 
>>> become
>>> more frequent, I ended up having to reinstall... This time I'm starting
>>> from the 151a3 disk, not the 148 one, and I'll avoid setting 
>>> esoteric paths
>>> and the like (after all, I kind of know where the software I need is 
>>> and I
>>> can call it by full pathname). The old Quadro was moved to my other
>>> machine, a dual boot Fedora16/WindowsXP rig I use for firmware 
>>> flashing and
>>> some light gaming.
>>>
>>> Thank you all.
>>>     Bryan
>>>
>> _______________________________________________
>> OpenIndiana-discuss mailing list
>> OpenIndiana-discuss at openindiana.org
>> http://openindiana.org/mailman/listinfo/openindiana-discuss
>