SPL disparity errors counter conflict

Elliott, Robert (Server Storage) Elliott at hp.com
Tue Apr 28 17:49:45 PDT 2009


Formatted message: <a href="http://www.t10.org/cgi-bin/ac.pl?t=r&f=r0904283_f.htm">HTML-formatted message</a>

The Protocol-Specific Port log page may contain both versions of the
counters:
a) It always contains the saturating versions in bytes 32-47 of the SAS phy
log descriptor;
b) It may also contain the wrapping versions in bytes 52-m of the SAS phy log
descriptor in the phy event descriptor, if the logical unit chooses to
implement those phy event sources (that selection is vendor-specific).
From: Penokie, George [mailto:George.Penokie at lsi.com]
Sent: Monday, April 27, 2009 3:05 PM
To: Penokie, George; Elliott, Robert (Server Storage); t10 at t10.org
Subject: RE: SPL disparity errors counter conflict
Rob,
OK, so everyone tells me there has to be two counters to make this work. But
that's not the issue.
The issue is that in section 4.11 The first paragraph states that the
Protocol-Specific Port log page has the counters as saturating and then a few
paragraphs down there is a statement+table that indicates that the
Protocol-Specific Port log page has the counters as wrapping. If you look in
the Protocol-Specific Port page the fields that hold the counters point to
the SMP REPORT PHY ERROR LOG function that states the counters are
saturating. It appear to me like the entry in the table description that
lists the Protocol-Specific Port log page is wrong and needs to be deleted.
If that is not the case then someone will have to explain to me how anyone
can figure out which if the two counters to place into the Protocol-Specific
Pot log page.
4.11 Phy events
Phys shall count the following events using saturating counters and report
them in the Protocol-Specific Port log page (see 9.2.8.1) and/or the SMP
REPORT PHY ERROR LOG function (see 9.4.3.11):
invalid dwords received;
dwords received with running disparity errors;
loss of dword synchronization; and
phy reset problems.
...
The phy event source field, defined in ?able 1, is used in the
Protocol-Specific Port log page (see 9.2.8.1), the REPORT PHY EVENT function
(see 9.4.3.14), the REPORT PHY EVENT LIST function (see 9.4.3.16), and the
CONFIGURE PHY EVENT function (see 9.4.3.30) and indicates or specifies the
type of phy event in the accompanying phy event field.
Bye for now,
George Penokie
LSI Corporation
3033 41st St. NW
Suite 100
Rochester, MN 55901
507-328-9017
george.penokie at lsi.com
________________________________
From: owner-t10 at t10.org [mailto:owner-t10 at t10.org] On Behalf Of Penokie,
George
Sent: Monday, April 27, 2009 8:26 AM
To: Elliott, Robert (Server Storage); t10 at t10.org
Subject: RE: SPL disparity errors counter conflict
How is it possible for one counter to have two diametrically opposed
characteristics? A single counter cannot be both wrapping and saturating at
the same time.
Bye for now,
George Penokie
LSI Corporation
3033 41st St. NW
Suite 100
Rochester, MN 55901
507-328-9017
george.penokie at lsi.com
________________________________
From: owner-t10 at t10.org [mailto:owner-t10 at t10.org] On Behalf Of Elliott,
Robert (Server Storage)
Sent: Tuesday, April 14, 2009 8:32 PM
To: t10 at t10.org
Subject: RE: SPL disparity errors counter conflict
Both.
When reported via the old REPORT PHY ERROR LOG function, they are always
saturating values.  They can be cleared with the PHY CONTROL function CLEAR
ERROR LOG phy operation, but that's not multi-initiator friendly.
When reported via the REPORT PHY EVENT LIST function (if those events have
been selected as sources - there are many other sources also available), they
are wrapping values.  These are multi-initiator friendly.
________________________________
From: owner-t10 at t10.org [mailto:owner-t10 at t10.org] On Behalf Of Penokie,
George
Sent: Tuesday, April 14, 2009 6:45 PM
To: t10 at t10.org
Subject: SPL disparity errors counter conflict
The running disparity errors counter in SPL (and SAS-2) in two places is
called out as a saturating counter and one place as a wrapping counter. (see
below for the wording) Note that WC stands for wrapping counter.
So which is it. It appears that the specification in table 37 (SPL r1) is not
correct, any comments.
4.11 Phy events
Phys shall count the following events using saturating counters and report
them in the Protocol-Specific Port log page (see 9.2.8.1) and/or the SMP
REPORT PHY ERROR LOG function (see 9.4.3.11):
a) invalid dwords received;
b) dwords received with running disparity errors;
c) loss of dword synchronization; and
d) phy reset problems.
Table 37 - PHY EVENT SOURCE field (part 1 of 4)
02h Running disparity error count	     WC        Number of dwords
containing running disparity errors (see 5.3.5) that have been received
outside of phy reset sequences
9.4.3.11 REPORT PHY ERROR LOG function
The RUNNING DISPARITY ERROR COUNT field indicates the number of dwords
containing running disparity errors (see 5.3.5) that have been received
outside of phy reset sequences. The count shall stop at the maximum value.
The RUNNING DISPARITY ERROR COUNT field is set to a vendor-specific value
after power on.
Bye for now,
George Penokie
LSI Corporation
3033 41st St. NW
Suite 100
Rochester, MN 55901
507-328-9017
george.penokie at lsi.com



More information about the T10 mailing list