Same problem, new year 2100A machine check

[Tom Ponsford]

Hi All,

I did think of trying to dig up a EISA configuration utility to see if that 
may be the problem. It did however boot and run quite well under 4.xx on a 
uniprocessor kernel, without any reconfiguration of the EISA bus. Under 4.xx 
it also did a machine check under a SMP kernel. Apparently, the 2100A is quite 
an evil machine with it's split PCI bus, I did remove all PCI cards from the 
machine, as there were reports a machine checks with certain cards. Removing 
the all the cards had no effect, however.

Bernd Walter wrote:
> On Sun, Jan 25, 2004 at 12:57:41PM +0100, Wilko Bulte wrote:
> 
>>On Sun, Jan 25, 2004 at 12:51:12PM +0100, Bernd Walter wrote:
>>
>>>On Fri, Jan 23, 2004 at 05:06:14PM -0700, Tom Ponsford wrote:
>>>
>>>>I decided to replace a disk on the ol' 2100A. I had been running VMS on it 
>>>>without any troubles, until a disk started giving me errors. After 
>>>>replacing the disk, I decided to give 5.2 a go, thinking maybe some of the 
>>>>problems of last year may have been resolved.
>>>>
>>>>No such luck.
>>
>>...
>>
>>
>>>>unexpected machine check:
>>>>
>>>>    mces    = 0x1
>>>>    vector  = 0x670
>>>>    param   = 0xfffffc0000006000
>>>>    pc      = 0xfffffc00003b48d4
>>>>    ra      = 0xfffffc00003b4908
>>>>    curproc = 0xfffffc0000810ee0
>>>>        pid = 0, comm = swapper
>>>>
>>>>panic: machine check
>>>>cpuid = 0;
>>>>Uptime: 1s
>>>>Automatic reboot in 15 seconds - press a key on the console to abort
>>>
>>>Maybe there is something wrong with the EISA configuration on your
>>>machine.
>>
>>Hmm, could be. But 2100[a] are not really well supported/tested..
> 
> 
> It's mostly guessing, because it's probing the bridge and EISA support
> is a difference between 4.x, which works acording to the originator.
>