[fm-discuss] Machine check exception not processed

[Mike Shapiro]

On Thu, Dec 13, 2007 at 02:44:20PM -0500, Matty wrote:
> Howdy,
> 
> We had a Sun X2200 running Solaris 10 update 3 panic with the following message:
> 
> Dec 12 14:37:46 bar cpu.generic: [ID 369736 kern.warning] WARNING:
> Machine-Check Exception in kernel mode

The X2200 with latest Opteron CPUs includes a significantly more
advanced cpu->kernel RAS interface, and requires new Solaris FMA code
to handle it.  When we did x64 FMA we made it such that Solaris
dynamically loads the right FMA code, or falls back to cpu.generic
if that code isn't there yet.  The FMA code for these latest Opteron
parts is in Solaris Nevada/OpenSolaris now and will be available
in the next Solaris 10 Update.

-Mike


> Dec 12 14:37:46 bar last message repeated 1 time
> Dec 12 14:37:46 bar cpu.generic: [ID 642271 kern.warning] WARNING: 2
> STAT 0xb60000000002018a ADDR 0x000000000000474b
> Dec 12 14:37:46 bar genunix: [ID 843051 kern.info] NOTICE:
> SUNW-MSG-ID: SUNOS-8000-0G, TYPE: Error, VER: 1, SEVERITY: Major
> Dec 12 14:37:46 bar cpu.generic: [ID 369736 kern.warning] WARNING:
> Machine-Check Exception in kernel mode
> Dec 12 14:37:46 bar cpu.generic: [ID 642271 kern.warning] WARNING: 2
> STAT 0xb60000000002017a ADDR 0x000000000000474b
> Dec 12 14:37:46 bar unix: [ID 836849 kern.notice]
> Dec 12 14:37:46 bar ^Mpanic[cpu2]/thread=ffffffffb9712440:
> Dec 12 14:37:46 bar genunix: [ID 663943 kern.notice] Unrecoverable
> Machine-Check Exception
> Dec 12 14:37:46 bar unix: [ID 100000 kern.notice]
> Dec 12 14:37:46 bar cpu.generic: [ID 369736 kern.warning] WARNING:
> Machine-Check Exception in kernel mode
> Dec 12 14:37:46 bar cpu.generic: [ID 642271 kern.warning] WARNING: 2
> STAT 0xf60000000002018a ADDR 0x000000000000474b
> Dec 12 14:37:46 bar unix: [ID 836849 kern.notice]
> Dec 12 14:37:46 bar ^Mpanic[cpu2]/thread=ffffffffb9712440:
> Dec 12 14:37:46 bar genunix: [ID 663943 kern.notice] Unrecoverable
> Machine-Check Exception
> Dec 12 14:37:47 bar unix: [ID 100000 kern.notice]
> Dec 12 14:37:47 bar cpu.generic: [ID 369736 kern.warning] WARNING:
> Machine-Check Exception in kernel mode
> Dec 12 14:37:47 bar cpu.generic: [ID 642271 kern.warning] WARNING: 2
> STAT 0xf60000000002011a ADDR 0x000000000000474b
> Dec 12 14:37:47 bar unix: [ID 836849 kern.notice]
> Dec 12 14:37:47 bar ^Mpanic[cpu2]/thread=ffffffffb9712440:
> Dec 12 14:37:47 bar genunix: [ID 663943 kern.notice] Unrecoverable
> Machine-Check Exception
> Dec 12 14:37:47 bar unix: [ID 100000 kern.notice]
> 
> Does anyone happen to know why FMA isn't able to diagnose whatever the
> underlying problem is? SUNOS-8000-0G doesn't really provide any
> details.
> 
> Thanks for any insight,
> - Ryan
> -- 
> UNIX Administrator
> http://prefetch.net
> _______________________________________________
> fm-discuss mailing list
> fm-discuss at opensolaris.org

-- 
Mike Shapiro, Solaris Kernel Development. blogs.sun.com/mws/