This message is from the T13 list server.

In just a few emails this afternoon we have defined the requirements
for a 

   >>>_*NEW*_<<<

feature in ATA hard disk drives. The requirement of the >_*NEW*_<
feature is that a drive keep a long term record of those sectors a
host has "Marked As Having Bad Data". I don't know what to call this
(bad, locked, come to mind but are not good terms). For now I will
just use MAHBD to describe just sectors. 

The basic requirment seems to be this: A host needs to be able to
select any sector and tell the drive to make it a MAHBD sector. The
sector will remain a MAHBD sector until it is successfully written by
the host or the MAHBD status is cleared by the host. Attempting to
read a sector marked MAHBD shall result in an UNC (uncorrectable data
error). In addition, MAHBD sectors shall not be affected or counted
by SMART or other drive media scanning activities.

Is this difficult to implement? Probably not. After all flash memory
ATA devices (CF for example) keep a record of which sectors are
currently "erased" and these devices have a command that returns the
current "erased" status of a sector.

I seems that we new a new command or commands to implement this
feature:

* a command to set MAHBD status for a sector.
* a command to clear the MAHBD status for a sector.
* a command to query the MAHBD status for a sector.
* maybe a command to clear all MAHBD status.
* maybe words in the ID data giving the maximum and current
  number of MAHBD sectors a drive supports.
* probably also need a command to "lock" the MAHBD commands against 
  improper use by a virus.

There are lots of issues to consider, mostly interactions with other
features: Host Protected Area, Security, SMART, etc. And then there
are the nasty issues in the tech support area that goes something
like this: "The drive was part of a RAID and never reported an error
but now it is full of sectors with uncorrectable errors that were
never reported by SMART. I want an RMA number and a new drive."

THIS IS A NEW FEATURE. Never before have I heard anyone say that a
drive needed to support the concept of a sector that the host wants
to contain "bad data". I have never seen a drive that explicitly
supported such a feature. Tradition is that a drive never has sectors
with "bad data". Much effort goes into making hardware and firmware
to insure a drive never has a sector with "bad data". 

Should I get a document number for this proposal or does someone else
want to take it on?



*** Hale Landis *** www.ata-atapi.com ***



Reply via email to