Subject: Re: sd2(ncr0:6:0): command aborted... ?
To: None <torppa@cute.fi>
From: Charles M. Hannum <mycroft@ai.mit.edu>
List: current-users
Date: 06/15/1995 09:42:14
   ncr0 targ 6 lun 0: <MICROP, 1936-21MW1002004, HW0A> SCSI2 0/direct fixed
   sd2 at scsibus0sd2(ncr0:6:0): asynchronous.
   sd2(ncr0:6:0): command aborted, data = 00 00 00 00 49 00 00 00 00 00
   : drive offline

According to the SCSI spec, that's:

49 00 DTLPWRSOMC INVALID MESSAGE ERROR

That's probably a bogus response to a sync negotiation request.  (SCSI
2 drives should be using MESSAGE MESSAGE REJECT.)  In
/sys/scsi/scsi_base.c, you could try:

  		case 0x8:	/* BLANK CHECK */
  			error = 0;
  			break;
+		case 0xb:	/* COMMAND ABORTED */
+			return ERESTART;
  		case 0xd:	/* VOLUME OVERFLOW */
  			error = ENOSPC;
  			break;

That will cause it to retry on a COMMAND ABORTED sense key, rather
than failing on the first try.

   >From that time syslog seems to have logged this before it crashed
   Jun 13 21:52:29 hauki /netbsd: sd2(ncr0:6:0):  DELAYED ERROR, key = 0x3
   Jun 13 21:52:29 hauki /netbsd: sd2(ncr0:6:0): medium error, info = 
   -860574551 (decimal), data = 00 00 00 00 12 00 00 20 10 00

That's:

12 00 D   W  O   ADDRESS MARK NOT FOUND FOR ID FIELD

It presumably indicates that there is a bad spot on your disk,
probably noticed by the drive during write-behind, after it told you
the data was written.