[openib-general] DAPL for openib

[Hal Rosenstock]

On Wed, 2004-08-25 at 17:25, Sean Hefty wrote:
> If you use separate tables for version, class, and method, and let them grow dynamically, 
> they can be substantially smaller.  E.g. the version table would likely be a single entry 
> referencing a class table.  

For most classes this is true. For CM, there might be a couple of
versions supported.

> The class table is about 8 entries long, 

This of course depends on what services are running on the node.

> but requires remapping class 0x81 to index 0.  

This remapping won't be done based on the recent email exchange of not
combining the DR/LR SM registration. (This is a minor alteration
to Sean's point about memory consumption.)

> The method array is only needed if a client doesn't register to 
> receive all unsolicited MADs for a specific class.

This means that for any class, the IB MAD layer needs to know
which methods are used for unsolicited MADs. 

By class, here are the potential unsolicited methods:
SM: trap
SA client: report
SA: all requests (get, set, gettable, getmulti, gettrace, delete)
CM: send
PM: none
BM: send, trap, report
DevMgt: trap, report
SNMP: send
vendor & application: send, trap

> Something like this should let you dispatch to a single client.  
> To support a client viewing the MAD, you'd just need linked lists at any point 
> along the dispatch path.