Subject: Completion of TRAIN or TRAIN_DONE sequences
Date: Tue, 19 Feb 2008 18:25:23 -0800
From: "Geoff Hibbert" <geoff.hibbert@finisar.com>
To: <t10@t10.org>
X-Message-Number: 8539
Formatted message: HTML-formatted message

Hello all,
I have a question regarding a possible ambiguity in the SAS-2 spec at
6.7.4.2.3.
It says:
"If the phy's receiver is trained and acquires dword synchronization
before TLT, then the phy shall stop transmitting TRAIN patterns and
start transmitting TRAIN_DONE patterns."
It also says that upon transmitting 4 or more TRAIN_DONE patterns and
receiving the minimum 1 TRAIN_DONE before MTT,
"then the phy shall:
a)	 stop transmitting TRAIN_DONE patterns;
b)	 start transmitting dwords from the link layer; and
c)	 consider the Train-SNW to be valid."
When it says "stop transmitting" is the intention that TRAIN and
TRAIN_DONE patterns always be sent in their entirety (primitive sequence
followed by 58 scrambled dwords) before switching to the next item, or
may the phy interrupt whatever TRAIN/TRAIN_DONE sequence it is in the
middle of transmitting to switch to transmitting the next item?
Thanks,
-Geoff Hibbert
-Finisar Corporation