[MtFuji] SATA cabcon proposal for Zero Power ODD effort
Leete, Brian A
brian.a.leete at intel.com
Wed Sep 2 11:32:08 PDT 2009
* From the T10 Reflector (t10 at t10.org), posted by:
* "Leete, Brian A" <brian.a.leete at intel.com>
I think we need to step back a bit and look at the big picture. What you and
others have raised is essentially the question of how we handle backwards
There are two fundamental problems with backwards compatibility 1) functional
compatibility, and 2) electrical compatability
Functional compatibility we have solved by the bit in IDENTIFY_PACKET_DEVICE. Electrical compatibility is a more tricky problem to solve.
An electrical compatibility problem can be solved 3 ways
1) Have the host protect itself from the drive
2) Evaluate the current state of the pin and hope that the current
electrical signaling of the drive is compatible enough to allow operation
3) Re-define the behavior of the drive (which is what I tried to do with
#1) I think requiring the host to electrically isolate the MD/Device
Attention pin would be very difficult. The additional cost could very well
prevent it's adoption. Also, I don't think it's a good long term solution
(short term may be OK) to have the host continually isolate the pin.
Any good architectural solution has pros and cons. One of the cons associated
with the two pin solution was the concept that the drive now has to behave
differently in a host system and in a manufacturing diagnostic environment.
This could potentially require modification of the drive manufacturing
process to put the drive in the manufacturing diagnostic mode. If this is
impossible, rather than have the host protect itself from the drive, the
better solution would be to go back to an earlier proposal where we re-use
the Device Present pin to provide this signaling. This would have no
electrical incompatibilities with the host system.
#2) Perhaps we could do a survey of what the drives are doing with this pin.
If a large majority of the drives can coexist electrically with the host
design as I have envisioned it, we can write this into the SATA spec and not
have the majority of the drives change significantly. Correct me if I'm
wrong, but RS232C is 5V signaling on that pin. There is probably going to be
a problem with feeding back a 5V TTL signal on that pin to the host.
#3) If we have to re-define the behavior of the drive, I'm open to other
suggestions than the 100K resistance number. We can ground it, we can let it
float, or whatever. It is reasonable that the host can control the
interpretation of the signal, so as long as there is no electrical damage to
the host or drive.
From: owner-mtfuji5 at avc-pioneer.com [mailto:owner-mtfuji5 at avc-pioneer.com] On
Behalf Of keiji_katata at post.pioneer.co.jp
Sent: Tuesday, September 01, 2009 11:36 PM
To: mtfuji5 at avc-pioneer.com
Cc: t10 at t10.org
Subject: Re: [MtFuji] SATA cabcon proposal for Zero Power ODD effort
Hi Brian, this is an additional two comments.
Comment #2 for legacy drive:
---- page 5 ----
Device implementation of this feature is optional and if supported, the
shall set the appropriate bit in the IDENTIFY PACKET DEVICE data structure to
inform the host it supports this capability. (See section 13.2 for details).
If the device doesn't implement the feature the device shall present a
resistance greater than 100 Kilohm (+/- 10%).
I think this is an optional feature, therefore the 2nd paragraph of above may
Pioneer's some legacy drive used MD pin as TTL level RS-232C output. So a
system that can support this feature should turn this feature off with legacy
Comment #3 for host setting at power-on state of drive:
For this purpose, the Device Attention signal should be detected by host at
power-off state of drive. If host has mask capability of DA pin of host, host
may stop the DA pin function of host at power-on state of drive.
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org
More information about the T10