This message is from the T13 list server.

Here are the T13 reflector items that have no clear resolution since the
June meeting.

Dan:  Please put these on the agenda.

All:  Please review your positions and be ready to comment.

-------------------------------
6/23  Joseph Chen       Read lookahead (ATA8-ACS)

Should ATA8-ACS rename this to SET FEATURES subcommand to 'enable/disable
read cache' ?

There was a lengthy discussion on the reflector, with no clear concensus.

------------------------------
6/30  Yamini Shastry          Count register (ATA8-ACS)

The description for the count field in ATA8-ACS for *Ext commands such as
Read Multiple Ext, Read Sectors Ext, Read Verify Sectors Ext etc., says:

"Count: The Number of sectors to be transferred. A value of 00h indicates
that 256 sectors are to be transferred."

Should that be 65536 instead of 256?
-----------------------------------
7/5   Alexander Krebs   PIO "DRQ data block' definition (ATA8-STP)

The definition is not 'crystal clear enough to implement PATA emulation on
SATA
with consistency and accuracy.

A lengthy discussion showed that there are several special cases.

Perhaps we need a 'note' in the 'emulation' section of the SATA document.
--------------------------------------
7/6   Yamini Shastry          LBA parameter in FIS (ATA8-AST)

In the  ATA8-AST spec: the description for  FIS fields such as those of
Response FIS refer to the ATA8-ACS spec.

For ex: The LBA field in the Response FIS in ATA8-AST sec 7.5.3.1 says that
the value depends on the type of command being sent and refers to the
ATA8-ACSxxx. Lets take the example of a WRITE DMA EXT (7.63), the ATA8-ACS
spec specifies that the Normal  Output (table 64) and Error Output (Table
78) for this command, have the LBA and count fields reserved. Now does that
mean that the Response FIS  from the device to host to the WRITE DMA EXT on
a serial transport should actually have these fields reserved (I.e. set to
zero)?

I ask is this because I always see that the devices actually set this to
the Ending LBA address for the I/O but I guess the receiver can't rely on
that. Now is it even expected that the device would set the LBA value on
the response when the ATA8-ACS spec says they are reserved?
----------------------------
7/7   Mark Overby       Question on text for SECURITY ERASE UNIT (ATA8-ACS)

In reviewing portions of ATA/ATAPI-7 relating to SECURITY ERASE UNIT, I
have found that there may be a conflict in the information that could lead
to unexpected behavior as observed by the host. I'm bringing this up here
so that the collective knowledge of the original intent of SECURITY ERASE
UNIT can be used to determine if clarification or other is needed.

If a host encounters a drive that has no password set and is currently in
the disabled / unfrozen state (state SEC1 from the security mode state
diagrams in figure 5, ATA/ATAPI-7 volume 1) the SECURITY ERASE UNIT (and
the requisite SECURITY ERASE PREPARE) command is allowed to be sent (Table
4). However, in the text for SECURITY ERASE UNIT, clause 6.44.8 it states:
“If the password does not match the previous password saved by the device,
the device shall reject the command with command aborted.”

What happens if no password has been set? Should the host issue a password
of all zeros? Does it matter what the host sends? Is it expected that this
should work? The behavior of when no password has been set is unclear.
----------------------------
7/12  Curtis Stevens          READ LOG EXT  (ATA8-ACS)

There is an inconsistency in ATA/ATAPI-7 that I think needs to be addressed
in the errata that is currently under development…

READ LOG EXT is not prohibited for ATAPI devices
WRITE LOG EXT is prohibited for ATAPI devices

There is no mention of GPL commands in the packet feature set description.
I believe that ATA/ATAPI-7 would be more consistent if the prohibited
statement were removed from the WRITE LOG EXT.  In ATA8 we could make WRITE
LOG EXT optional??
--------------------------------------
7/13  Mark Overby       FLUSH CACHE EXT and FLUSH CACHE behavior (ATA8-ACS)

What is the behavior of a drive if FLUSH CACHE EXT and FLUSH CACHE if the
write cache is disabled. Under the current reading of ATA/ATAPI-7 (and 8),
the standard is silent. It would be legal for a device to abort the command
from my reading. Based on the text in the error outputs, it would appear
that the intent was that the device completes the command successfully even
if the write cache is disabled. This reading is based on the fact that the
described error case refers to the device being unable to write to the
media.

Is this the correct intent? If so, I suggest we propose language to make
that clear.

This comes up as a result of the SAT work in T10 with the behavior of START
STOP UNIT.
----------------------------------------


Thank You !!!
-----------------------------------------------------------------
Jim Hatfield
Seagate Technology LLC
   e-mail:  [EMAIL PROTECTED]
   s-mail:  389 Disc Drive;  Longmont, CO 80503 USA
   voice:  720-684-2120
   fax....:  720-684-2711
==========================================

Reply via email to