Re: Weird RAID/SATA problem [ once was Re: 2.6.17-mm3 ]

[Neil Brown <neilb@suse.de>]

On Sunday July 2, reuben-lkml@reub.net wrote:
> BUG: warning at drivers/md/md.c:411/super_written_barrier()
> 
> Call Trace:
>   <IRQ> [<ffffffff80422231>] super_written_barrier+0x61/0x100
>   [<ffffffff8023c000>] bio_endio+0x5a/0x6a
>   [<ffffffff8022e24f>] __end_that_request_first+0x16f/0x4c9
>   [<ffffffff8024afaa>] end_that_request_first+0xc/0xe
>   [<ffffffff8034e825>] blk_ordered_complete_seq+0x7d/0x8c
>   [<ffffffff8034e864>] post_flush_end_io+0x30/0x35
>   [<ffffffff8034e748>] end_that_request_last+0xd8/0xf4
>   [<ffffffff803d83a1>] scsi_end_request+0xb1/0xdd
>   [<ffffffff803d87cd>] scsi_io_completion+0x3cd/0x3dc

I think the key decision to return an error is happening here in
scsi_io_completion. 
Pooring over a disassembly might help show here, but sticking in a
bunch of printks is probably easier (for someone like me who has never
seen this code before anyway :-)

What does this patch produce?

NeilBrown

Signed-off-by: Neil Brown <neilb@suse.de>


diff .prev/drivers/scsi/scsi_lib.c ./drivers/scsi/scsi_lib.c
--- .prev/drivers/scsi/scsi_lib.c	2006-07-01 23:22:46.000000000 +1000
+++ ./drivers/scsi/scsi_lib.c	2006-07-01 23:26:18.000000000 +1000
@@ -952,6 +952,7 @@ void scsi_io_completion(struct scsi_cmnd
 				 * and quietly refuse further access.
 				 */
 				cmd->device->changed = 1;
+				printk("Unit Attention\n");
 				scsi_end_request(cmd, 0, this_count, 1);
 				return;
 			} else {
@@ -984,6 +985,8 @@ void scsi_io_completion(struct scsi_cmnd
 				scsi_requeue_command(q, cmd);
 				return;
 			} else {
+				printk("Illegal Request %d %d %d\n",
+				       sshdr.asc, sshdr.ascq, cmd->cmnd[0]);
 				scsi_end_request(cmd, 0, this_count, 1);
 				return;
 			}
@@ -1012,6 +1015,7 @@ void scsi_io_completion(struct scsi_cmnd
 					    "Device not ready: ");
 				scsi_print_sense_hdr("", &sshdr);
 			}
+			printk("Not ready\n");
 			scsi_end_request(cmd, 0, this_count, 1);
 			return;
 		case VOLUME_OVERFLOW:
@@ -1022,6 +1026,7 @@ void scsi_io_completion(struct scsi_cmnd
 				scsi_print_sense("", cmd);
 			}
 			/* See SSC3rXX or current. */
+			printk("Volume Overflow\n");
 			scsi_end_request(cmd, 0, this_count, 1);
 			return;
 		default:
@@ -1045,6 +1050,10 @@ void scsi_io_completion(struct scsi_cmnd
 				scsi_print_sense("", cmd);
 		}
 	}
+	printk("ouch. %d %d %d   %d %d %d  %d\n",
+	       good_bytes, sense_valid, sense_deferred,
+	       sshdr.sense_key, sshdr.asc, sshdr.ascq,
+	       result);
 	scsi_end_request(cmd, 0, this_count, !result);
 }
 EXPORT_SYMBOL(scsi_io_completion);
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/