Toll Free Dial in Number: (866) 279-4742
International Access/Caller Paid Dial In
Number: (309) 229-0118
PARTICIPANT CODE: 3243413
Webex information:
https://seagate.webex.com/seagate Topic: SAS-2 PHY WG
Date: Thursday, July 20, 2006
Time: 10:00 am, Central Daylight Time (GMT
-05:00, Chicago)
Meeting number: 826 515 680
Meeting password: 6gbpsSAS
Agenda: Speed negotiation sequence
During the face-to-ace meeting in Colorado
Springs, a new proposal for the G3 window was introduced that uses an out
of band communication to broadcast PHY capabilities. This method preserves
the existing speed negotiation window structure without depending on G1
or G2 capability to work in future generations. I have provided a preliminary
list of considerations concerning 05-397r5 in the meeting minutes and have
included them below. There has also been a proposal posted on the T10 site
concerning how the out of band data could be configured. We will discuss
these proposals and concerns during the call (limited to approximately
2 hours)
Uses OOB at 1.5Gbps with a 9uS idle and a
10uS initial burst to "start the clock". The 59 10uS intervals
start from the far end of the 609uS standard window. Later this was changed
to a 10us idle at the end and 58 intervals for data.
Since OOB already requires the 1.5Gbps bursts,
this proposal has a subset of the existing OOB requirement and is actually
less demanding because of the relatively long signal bursts.
This OOB section has been moved to the existing
G3 window following the G2 window rather than having “G2 supporting SSC”
in the G3 slot. After this G3 window, the final speed negotiation window
is entered at the highest common feature set enabled.
One suggestion was to make the intervals
a binary code with a table to decode what it means.
Present method uses the particular time area
as positive indication of support if active signal.
The intervals may need to include an idle
time if the squelch detector is used, as a discharge time may be required
to accurately detect the presence of signal.
Failure of the final speed negotiation window
shall be handled by a higher-level layer (not the PHY).
If there was a failure at the highest negotiated
speed and the initiator/expander changed the supported options, then how
would it reset to full capability if the device was changed?
A concern was raised that using the scrambler
in the training sequence may involve the link layer.
Since the last interval in the configuration
window is idle, the training data may start at the beginning of the final
speed negotiation window, but shall start by the end of a defined RCDT
(not necessarily the same length of time as the previous RCDT’s). Input
is needed on how long this RCDT should be.
How is the final speed negotiation window
completed? Should there be ALIGN0/ALIGN1 after TRAINdone is exchanged to
verify dword sync?
Reference documents:
SAS-2 Start-up training sequence (05-397)
[Newman]
http://www.t10.org/ftp/t10/document.05/05-397r5.pdf
Proposal New Speed Negotiation
for SAS-2 (06-354) [Finch]