kernel panic(?) trying to copy data off failed drive with dd

[Greg Lane]

On Mon, Jun 12, 2006 at 06:47:56AM -0400, "Michael P. Soulier" <msoulier at digitaltorque.ca> wrote:
> On 12/06/06 Miroslav Lachman said:
> 
> Regardless, file a bug report. The box should never hang, or reboot. 

Is this the case?  I had another box (a piece of crap I played around on) 
that had a failing disk, and this would bring the machine down. So I didn't 
think this was necessarily unusual.  However, on the previous occasion the disk 
was the root file system with swap on it, whereas in this case, the disk is 
only a data disk, not any part of the OS.  

Are you saying that the box should never hang or reboot, but should 
recover from the error, and whatever command was running should fail and 
return an error message?

If that is the case, is there possibly some sysctl or boot setting that 
I might have set that would cause it to fail catastrophically rather than 
properly.  Maybe disabling DMA would help? Something? Anything?

If people think this is abnormal enough to warrant a bug report, I am 
happy to file one. I want to be sure this is unexpected behaviour though. 

Some terse info here, in case it helps:

5.5-PRERELEASE from Wed May 10

Tyan Thunder K8S 2880 UGNR  (onboard vga, mpt scsi, dual bge ethernet)
4 x 1GB DDR400
2 x Opteron 248
1 x 73 GB SCSI (Maxtor)   (/ /usr etc, plus 1 x 6 GB swap on here)
1 x 147 GB SCSI (Maxtor)  (/home,  plus 1 x 6 GB swap on here)
2 x SATA 250GB (WD)  (data disks, it is one of these that has failed)
1 x ATA 250GB (WD)   (scratch disk)
Adaptec 2940 with a DLT and EXABYTE drive

I can send a dmesg directly if anyone has any ideas.

Greg