home *** CD-ROM | disk | FTP | other *** search
-
- The October Unix CAM OSD meeting was held in Orlando Florida
- on 10/6/89. The following people were in attendance at this meeting:
-
-
- Steve Betts (NCR
- Joe Hinrichs (NCR
- Joe Lawlor (AT&T
- David O'Shea (Future Domain
- Darryl Richman (Interactive Systems
- Janet Schank (DEC
- Bob Snively (Sun
- Eric Tausheck (HP
-
- Topic that were discussed were as follows
-
- A document generated by Tom Shea and Dal Allen that came through
- the mail in the last month was discussed. This is a proposed first
- draft of the CAM "working document". Most of the Unix OSD was dedicated
- to evaluating Secion 8.1 (CAM Control Block Definition). The following
- are some of the issues that came out of this discussion:
-
- - Defining pointer size (in bytes) is hopeless in the Unix
- world due to the diversity of the platforms that Unix runs
- on. It was recommended that some words be added to the working
- document on the order of "For non-Intel based computers, the
- size of the pointers in the above structures may vary ...
-
- - It was recommended that due to the dynamic nature of the CAM
- structure at this time (due to design changes), no significant
- attempt should be made to align elements in the structure.
- However, a note below the structure should point out that
- "Note: reordering is likely ... alignment will be worked out
- in the future when the structure is more stable".
-
- - Section 8.1.3 "CAM Status" and "HBA Status" should be
- integrated into one return field. It is not always clear
- how to differentiate these two fields and performance can
- possibly be enhanced by having one field.
-
- - Section 8.1.3 "CAM Status" should have the following additional
- return codes:
- a) CAM Busy - cannot take any more requests at this time
- b) Timeout - timeout value provided by the target driver
- has now expired
- c) Unexpected bus free
- d) (possibly others from appendix D
-
- - Section 8.1.5 (CAM Flags) - The data direction field should not
- have a code "00". Anyone issuing a CCB should (better) be able
- to determine the direction of the data.
-
- An issue was raised (S. Betts) concerning using "Streams" format Unix
- drivers. The question of performance issues differences between a streams
- format driver and a "stock" driver was never resolved(at least to my
- satisfaction). However, the consensus was to avoid Streams format because
- it would require Target Drivers as well as Host Adapter drivers to adhere
- to the interface, which would presumably make porting more difficult
- across Operating Systems.
-
- Steve Betts provided a proposal that defined an Device Equipage Tabl
- that would be administered by the CAM Router. A psudo-coded scan
- algorithm was also provided. Time did not allow for any major discussion
- on this proposal.
-
-
- =========================================================================
- NEXT MEETING
- The Next SCSI CAM meeting is in Santa Ana Ca
- It is sponsored by Future Domain (Ken Post (714) 259-0400)
-
- Meetings will be at the Embassy Suites Hotel (714) 241-3800
- Reservation Cutoff - October 13, 1989
- SCSI-3 Working Group October 30 - 3
- Unix CAM OSD Group October 31 at 1:00 PM
- CAM Working Group November 1 at 9:00 A
-
- **** Place of Unix OSD has not been determined. This issue is being
- **** worked by Ken Post. As soon as I get this info, I'll post/email.
- =========================================================================
-
- Since our meeting time is so limited, if you have a proposal for the next
- meeting, please try to email or FAX it out before the meeting. This will
- give people a chance to get some understanding of the material.
- Here are some agenda items for the October 31st meeting. I can send out
- a more formal agenda when I get feeback on proposals and areas of
- concern.
- 1) Further review of the CAM "working document". Dal should be
- sending a new revision sometime before the next meeting
-
- 2) Initialization and configuration. We need to evaluate what the
- Unix needs are for these functions. What needs to be
- added/changed in the "working document" and the "Steve Betts
- Initialization" document to fulfill these needs.
-
- ==========================================================================
-
-
- Joe Lawlor
- AT&T Bell Lab
- (312) 979-0133
- lawlor@iwtio.att.com
-