Remapping bad sectors (was: Problems with aic7880 (AGAIN))

Mike Bilow mikebw at bilow.bilow.uu.ids.net
Mon Mar 30 07:56:52 PST 1998



Doug Ledford wrote in a message to Mike Bilow:

 DL> Actually, you are correct that verify media should remap bad
 DL> sectors if the ARRE bit is set.  However, not to point at
 DL> drive manufacturers, some of the recent drive firmwares I've
 DL> dealt with have essentially acted as though the AWRE/ARRE
 DL> bits are unset regardless of whether they are set or not. 
 DL> Furthermore, on those drives, if you ever do actually
 DL> receive a media error from the drive, then once it's made
 DL> its way to the front where you can see it, the drive will
 DL> *never* remap that sector until a low level format is done. 
 DL> So in short, if all drives act like they are suppossed to,
 DL> then you are 100% correct in your email, but because of a
 DL> few naughty boys, I always go the safe route and recommend
 DL> the low level format. 

A simple test is to run "Verify Media" twice, which is good practice anyway. 
If any bad sectors are remapped on the first pass, you will get a red text
message box giving the sector address and be asked to press a key to authorize
the remapping.  You should always make notes of the sector addresses remapped. 
The drive should be fully powered down and restarted after completing every
low-level format or verify pass.

On the second pass, you should expect a flawless drive.  If any errors are
encountered, compare the sector addresses reported.  If they are the same
sectors on the first and second passes, then obviously remapping is not
working.  If they are different sectors, then it is possible that the drive is
simply going bad, often due to spindle wobble or motor speed control problems.
Cabling problems can also account for "bad spots" that appear to move randomly.

It is also possible that remapping, either explicit or implicit, will fail if
there are already enough sectors remapped so as to consume the supply of
spares.  Such a drive should be replaced.

Many SCSI controllers write information to the drive with MODE SELECT when they
perform low-level format or verify operations.  If a drive is low-level
formatted on one brand of controller and then plugged into another, the
settings internal to the drive may not be as expected by the controller and
this can lead to unexpected operational problems.
 
-- Mike



To Unsubscribe: send mail to majordomo at FreeBSD.org
with "unsubscribe aic7xxx" in the body of the message



More information about the aic7xxx mailing list