System Issues - an update

Duncan.Penman at Duncan.Penman at
Fri Feb 24 22:33:48 PST 1995

This is a general update on the progress of the System Issues Study Group.  
I'll send out specific mail to each reflector next week, before the 
March X3T10 meeting.

Timetable: All issues, topics, problems, whatever, identified by March 3rd.
           System Issues Study Group meeting Wednesday morning, March 8,
           at the Newport Beach X3T10 meeting.  NOTE: we will review all
           the issues raised to date, but THIS MEETING IS ABOUT PRIORITIES
           AND PROCESSES (i.e. where and when issues get worked on, and who
           Recommendations from this meeting will be presented to the
           plenary session on Thursday.  New X3T10 projects for specific 
           issues may be generated there.

           The outcome of the Wednesday and Thursday meetings will be 
           presented to the SFF meeting Friday morning.  This will probably
           be tied in with a discussion of Dal Allan's proposal to
           restructure SFF into 3 different focus areas: connectors, host
           hardware, and host software.


     1. Assignment of drive letters by the OS is not consistent.  Seems 
        to be restricted to the PC platform.  This is not uniquely a SCSI
        issue but that is the context in which it was first presented.  

     1. Need a definition of minimum requirements to be an
        "ATAPI-compliant" drive.  Possibly a test suite to verify it.
     2. ATASPI.  Is it needed?  Should it be an X3T10 standard?  If so,
        what about ASPI?  At least 1 or 2 strong opinions have been
        expressed on this one!

     1. Large drive addressing.
     2. BIOS INT13-to-OS interface.
     3. I/O Port Address usage, both task file addresses for additional
        adapters and PCI bridge port addresses for IDE configuration
        and timing.
     4. (one I propose adding) A model of "HOST" structure and behavior
        needs to be developed.  It's painfully clear in many ATA
        discussions that people are talking at cross-purposes because
        (for example) one person is describing OS behavior and another is
        describing a local bus adapter chip function, yet both are saying,
        "The host has to ....".  No wonder it's difficult to agree.  We 
        need to expand our collective vocabulary and an explicit model
        is a good way to do it.

        I suggest we initially assume a PC platform, but PowerPC isn't 
        far off and some of the things 'we all know' might change.  A
        model could help there too.

     X. The ATA list isn't complete.  I'll put out a longer one next 
        week but can guarantee it won't be complete either.  My sense of 
        the direction for system issues is that ATA needs some organized, 
        long term activity.  ATAPI can ride on its coat tails.  SCSI is doing
        fine within the existing organization.

        Come to the system issues meeting on 3/8 or post your comments 
        here if you want to express yourself on this subject.


* Duncan Penman                         *
* IIX Consulting/ENDL Associates        *
* penman at                     *
* ph: (408)730-2565, fax: (408)730-5527 *

More information about the T10 mailing list