On 11/10/2011 10:44 AM, James Bottomley wrote:
On Thu, 2011-11-10 at 16:18 +0100, Hannes Reinecke wrote:
On 11/10/2011 03:55 PM, James Bottomley wrote:
On Thu, 2011-11-10 at 15:12 +0100, Hannes Reinecke wrote:
scsi_io_completion() tries to take some action based on
the command result and sense code. It also displays
'unhandled error code' or 'unhandled sense code' in case
no special handling was found.
This serves as an additional source of confusion to
the unsuspecting user, as the message in fact means
'everything okay, no special casing required',
and not 'oh gosh, something has happened and the system
couldn't deal with it'.
It means we're just about to fail the command with an error. That's not
really an everything ok case.
Sometimes, you're right, this is the correct thing to do silently, but
often it's not. So, same question to you as to Rob: What are the
circumstances you want silent failure for and can you special case them?
This is what I have seen (recent rhel6):
Nov 10 03:42:06 <host-name> kernel: sd 2:0:0:1: [sdb] Unhandled error code
Nov 10 03:42:06 <host-name> kernel: sd 2:0:0:1: [sdb] Result:
hostbyte=DID_NO_CONNECT driverbyte=DRIVER_OK
Nov 10 03:42:06 <host-name> kernel: sd 2:0:0:1: [sdb] CDB: Read(10): 28
00 00 4e 91 e0 00 00 20 00
Nov 10 03:42:06 <host-name> kernel: end_request: I/O error, dev sdb,
sector 5149152
Would the correct solution for this case to be to make description:
""
or a special case, something like:
"Couldn't connect before timeout period"
as I posted here:
http://marc.info/?l=linux-scsi&m=131854148309464&w=2
In the post I made above, I also liked
+ description = "Default sense code handling";
for the sense case.
and potentially something like
+ description = "Default host byte handling";
for the host_byte case when a host_byte isn't handled as
an alternative catch-all.
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
