Notify primitive issues

Elliott, Robert (Server Storage) Elliott at hp.com
Wed Jun 25 08:42:03 PDT 2003


* From the T10 Reflector (t10 at t10.org), posted by:
* "Elliott, Robert (Server Storage)" <Elliott at hp.com>
*

> The following are all issues relating to the NOTIFY primitive 
> that need to be addressed in either SAS or SAS 1.1. My 
> preference is SAS since it is going to have to go  through 
> another public review. 
> 
> 1-SAS currently has no statement or requirement for a target 
> to detect the Notify primitive. This could lead to 
> inoperability problems if the device sending the Notify 
> assumes the target will detect every Notify transmitted. It 
> is impossible to guarantee the detection of a single 
> primitive as it could be morphed in transmission causing a 
> disparity error. 
> I would recommend  the following be placed in section 7.2.5.9 NOTIFY: 
> 
> SAS initiator devices and expander devices shall not assume 
> that a SAS target device detects every transmitted NOTIFY.

OK.

> 2-Section 10.2.8.1.7.2 Transition SA_PC_5:Active_Wait to 
> SA_PC_1:Active should be changed from: 
> 
> This transition shall occur if: 
> 
> a) a NOTIFY (ENABLE SPINUP) is received; or 
> b) the SAS device does not consume additional power as a 
> result of the transition to SA_PC_1:Active. 
> 
> to: 
> 
> This transition shall occur if: 
> 
> a) a NOTIFY (ENABLE SPINUP) is detected; or 
> b) the SAS device does not consume additional power as a 
> result of the transition to SA_PC_1:Active. 

If it doesn't get through, it's just not received.  "Detected"
is a special term used for OOB signals.

> 3-Section 10.2.8.1.7.5 Transition SA_PC_5:Active_Wait to 
> SA_PC_6:Idle_Wait should be changed from 
> 
> This transition shall occur if: 
> a) a START STOP UNIT command with the POWER CONDITION field 
> set to IDLE is received; 
> b) a START STOP UNIT command with the POWER CONDITION field 
> set to FORCE_IDLE_0 is received; or 
> c) the Power Condition mode page idle condition timer expires. 
> For transitions based on a START STOP UNIT command, the 
> command shall not complete with GOOD status until this state 
> machine reaches the SA_PC_2:Idle state. 
> 
> to: 
> 
> This transition shall occur if: 
> a) a START STOP UNIT command with the POWER CONDITION field 
> set to IDLE is received; 
> b) a START STOP UNIT command with the POWER CONDITION field 
> set to FORCE_IDLE_0 is received; or 
> c) the Power Condition mode page idle condition timer expires. 
> For transitions based on a START STOP UNIT command, if the 
> IMMED bit is set to zero the command shall not complete with 
> GOOD status until this state machine reaches the SA_PC_2:Idle 
> state. For transitions based on a START STOP UNIT command, if 
> the IMMED bit is set to one the command may complete with 
> GOOD status before this state machine reaches the SA_PC_2:Idle state. 

OK.  (Perhaps similar text in SBC-2's IMMED bit description would
also be appropriate?)

> 4-Section 7.2.5.9 NOTIFY the statement: 
> 
> A specific NOTIFY shall not be transmitted a second time 
> until at least three ALIGNs or different NOTIFYs have been 
> transmitted. 
> 
> Allows NOTIFYs to completely replace aligns. This is OK as 
> long as the same NOTIFY is not the only NOTIFY being 
> transmitted.

It says "or different NOTIFYs".  If you're only sending one
NOTIFY, you aren't going to send 3 different ones and will
have to cycle through ALIGNs.

> To prevent this the wording should be changed to: 
> 
> A specific NOTIFY shall not be transmitted a second time 
> until at least three ALIGNs or three different NOTIFYs have 
> been transmitted. 

I don't see any functional difference in your wording.  If
anything, this would prevent ALIGN NOTIFY(b) NOTIFY(c) from
qualifying as allowing NOTIFY(a) to be sent.  

> 
> 
> Bye for now,
> George Penokie
> 
> Dept 2C6  114-2 N212
> E-Mail:    gop at us.ibm.com
> Internal:  553-5208
> External: 507-253-5208   FAX: 507-253-2880

--
Rob Elliott, elliott at hp.com
Hewlett-Packard Industry Standard Server Storage Advanced Technology
https://ecardfile.com/id/RobElliott


*
* 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 mailing list