[ofa-general] [PATCH] OpenSM/osm_port_info_rcv.c: In __osm_pi_rcv_process_endport, isSMdisabled also indicates that an SM is present so poll SMInfo

[Hal Rosenstock]

On Thu, 2007-04-05 at 14:16, Roland Dreier wrote:
>  > p.865 C14-53 and C14-54.1.1 state the behavior I originally said (an not
>  > active SM responds to SMInfo gets/sets). I think this superceeds the
>  > first bullet in C14-70 which says incoming SMInfos are dropped.
> 
> That's something else -- it's talking about a running SM that is in
> the NOT-ACTIVE state, because the master SM disabled it via a
> SubnSet(SMInfo).  But that wouldn't affect the IsSMDisabled bit, which
> is something different:

I put the two things together. Maybe that is wrong.

>   C14-69: If a SM can reside on a port, a vendor defined, out-of-band
>   mechanism shall be provided that when asserted will disable the
>   capability of running a SM from that port and the state of the
>   mechanism shall be indicated in the Portinfo:CapabilityMask.IsSMdisabled
>   bit.
> 
> So if IsSMDisabled then an SM is forbidden from running at all.  And
> I'm still confused -- why would anyone care whether a port has no SM
> running (ie IsSM is not asserted), or _really_ has no SM running (IsSM
> not asserted and IsSMDisabled asserted)?

Good point. At a minimum, the spec is unclear about this (if they are
totally separate mechanisms).

-- Hal

>  - R.