Proposal to add a new modifier to SCSI Write commands.

Tom coughlan at star.zko.dec.com
Wed Nov 27 12:20:55 PST 1996


* From the SCSI Reflector (scsi at symbios.com), posted by:
* coughlan at star.zko.dec.com (Tom)
*
Response to:

>* "Joseph Glider/Almaden/IBM" <gliderj at almaden.ibm.com>
>*
>The problem with implementing a 'forced error' function that simply shows up as
>an ECC error is that automatic reassignment algorithms are likely to end up
>reassigning blocks that are perfectly good.

The host gets an unrecovered read error, so it will not reassign the block.  

I agree that the algorithms in the drive must avoid reassigning good
blocks, but that mechanism is entirely the drive's business.  It does not
need to be reflected in a different status to the host.

>Any kind of 'forced error' implementation needs to include the capability of
>seeing that the resulting error is 'logically bad' rather than physically bad.

That may be true inside the drive, but they need not be distinguishable
externally.

>Does the DEC patent entirely cover the concept of 'logically bad' or just using
>the mechanism of using the ones complement of CRC to signal logically bad?

The DIGITAL patent:

	- Says that writing a block with Forced Error returns a 
	  specific error code when that block is subsequently read.

	- Requires that a block that returns Forced Error on a read can be
	  counted on to contain a correct copy of the data written to it
	  with the Forced Error flag. 

	- Describes an implementation that uses the ones complement of an
	  error detecting code to signal the forced error.

The proposal does not infringe, because it does not contain the first two
items, and does not require the third.  (If anyone is interested, the
patent is number 4,434,487, an overview of which can be obtained from:

  http://fetch.cnidr.org/cgi-bin/linker3?/pto7/NEW/INDEX+4434487+F )

To answer your question, I believe that the DEC patent does not entirely
cover the concept of 'logically bad'.  It also appears to me as though the
removal of the first two attributes listed above provides plenty of
implementation freedom.
*
* For SCSI Reflector information, send a message with
* 'info scsi' (no quotes) in the message body to majordomo at symbios.com




More information about the T10 mailing list