Do you have a preference?
 
We generally don't require an expander to snoop dwords inside a connection (except for certain primitives like CLOSE, BREAK, SATA_HOLD, SATA_HOLDA, and ERROR) so frame transmissions (SOF, SATA_SOF) and CREDIT_BLOCKED should probably not be counted.
 
These don't care about originated/destined vs. forwarded:
01h Invalid dword count (low-level)
02h Running disparity count (low-level)
03h Loss of dword synchronization count (low-level)
04h Phy reset problem count (low-level)
05h Elasticity buffer overflow count (low-level)
06h Received ERROR count (low-level)
25h Received AIP (WAITING ON PARTIAL) count (stated)
26h Received AIP (WAITING ON CONNECTION) count (stated)
2Ah Connection count
2Bh Peak transmitted pathway blocked count
2Ch Peak transmitted arbitration wait time
2Dh Peak arbitration time
2Eh Peak connection time
52h SATA flow control buffer overflow count
 
(2Ah-2Eh are related to connection establishment, which is not purely "forwarded")
 
These count originated/destined only, just because of their nature:
20h Received address frame error count (address frames are not passed along if there is an error)
27h Transmitted BREAK count  (BREAKs are never forwarded)
28h Received BREAK count (BREAKs are never forwarded)
29h Break Timeout count (BREAKs are never forwarded)
 
These count originated/destined only:
21h Transmitted abandon-class OPEN_REJECT count (stated)
22h Received abandon-class OPEN_REJECT count (stated)
23h Transmitted retry-class OPEN_REJECT count (stated)
24h Received retry-class OPEN_REJECT count (stated)
40h Transmitted SSP frame count
41h Received SSP frame count
42h Transmitted SSP frame error count
43h Received SSP frame error count
44h Transmitted CREDIT_BLOCKED count
45h Received CREDIT_BLOCKED count
50h Transmitted SATA frame count
51h Received SATA frame count
60h Transmitted SMP frame count
61h Received SMP frame count
63h Received SMP frame error count
 
A shared footnote for the last set is probably the best way to document this.

--
Rob Elliott, elliott@hp.com
Hewlett-Packard Industry Standard Server Storage Advanced Technology


 


From: owner-t10@t10.org [mailto:owner-t10@t10.org] On Behalf Of Bill.Martin@emulex.com
Sent: Thursday, November 15, 2007 1:06 PM
To: t10@t10.org
Subject: clarification of meaning of transmitted in phy events

In SAS2r12 table 37, there are a number of events that relate to received and transmitted primitives.  Some of these e.g., code 23h Transmitted retry-class OPEN_REJECT count, and code 25h Received AIP (WAITING ON PARTIAL) count, explicitly call out whether and expander shall count forwarded versions of these primitives.  The following PHY Events do not explicitly state what expanders shall do when forwarding these primitives:

 

40h

Transmitted SSP frame count

WC

Number of SSP frames transmitted.

41h

Received SSP frame count

WC

Number of SSP frames received.

44h

Transmitted CREDIT_BLOCKED count

WC

Number of times the phy transmitted a CREDIT_BLOCKED.

45h

Received CREDIT_BLOCKED count

WC

Number of times the phy received a CREDIT_BLOCKED.

50h

Transmitted SATA frame count

WC

Number of STP or SATA frames transmitted.

51h

Received SATA frame count

WC

Number of STP or SATA frames received.

60h

Transmitted SMP frame count

WC

Number of SMP frames transmitted.

61h

Received SMP frame count

WC

Number of SMP frames received.

 

 

Bill Martin

Emulex
Office of Technology
Industry Standards
916 772-3658
916 765-6875 (Cell)
bill.martin@emulex.com