about the SAS XL state machine
hamasaku at d1.hw.necst.nec.co.jp
Tue Dec 9 16:45:05 PST 2003
* From the T10 Reflector (t10 at t10.org), posted by:
* "Shogo Hamasaku" <hamasaku at d1.hw.necst.nec.co.jp>
Dear Elliott and everyone,
Even if we don't concern the case of the racing between
transitions and messages, the racing between Transmit Open
and the Request Path happens. Is it same (doesn't concerned
in the standard)?
If so, do you think it should be up to HW designer that
to add some ideas(insert the new state or new transitions?)
to prevent the deadlock?
Transmit Open indi.(from another XL) ________/~~~~~~~
Request Path req.(to ECM) ________/~~~~~~~
XL state (phy [Y]) _XL0__||__XL1___
NEC System Technologies, Ltd.
> * From the T10 Reflector (t10 at t10.org), posted by:
> * "Elliott, Robert (Server Storage)" <elliott at hp.com>
> > I was thinking about a motion of XL state machine with another
> > situation. I attached a pdf file that explains my situation.
> > What do you
> > think of a situation when the ECM receives the Request Path after, or
> > same time, sending the Transmit Open?
> > 1. An end device [A] sends the OPEN address frame to an
> > expander phy [X].
> > 2. The expander phy [X] receives Arb Won and sends the Transmit Open
> > to an expander phy [Y].
> > 3. An end device [B] sends the OPEN address frame to the expander
> > phy [Y].
> > 4. The expander phy [Y] happens following events at same time.
> > 4a. It receives the Transmit Open from the expander phy [X], and
> > 4b. It goes to XL1 state.
> > In above sequence, the ECM and the expander phy [X] could not stop
> > the Transmit Open for the expander phy [Y], and the expander phy [Y]
> > could not stop going to XL1 state.
> > And I thought an outgoing frame should be sent by expander phy [Y]
> > after comparing the outgoing frame and an inbound frame.
> As far as the standard is concerned (this is different from real
> hardware), transitions and messages do not happen at the exact
> same time. The message is always before or after the transition.
> If it happens before, then the state machine is still in XL0
> and makes the decision to handle the Transmit Open first
> (going to XL5).
> If it happens after, then the state machine sent Request Path
> from XL0 and entered XL1. The ECM should not send a Transmit
> Open because it has already received a Request Path. It must
> send an Arb Lost or Arb Reject before being allowed to
> send a Transmit Open after that time.
> Rob Elliott, elliott at hp.com
> Hewlett-Packard Industry Standard Server Storage Advanced Technology
> * For T10 Reflector information, send a message with
> * 'info t10' (no quotes) in the message body to majordomo at t10.org
* 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