On 6/2/05, Libor Michalek <[EMAIL PROTECTED]> wrote:
> On Thu, Jun 02, 2005 at 03:43:47PM -0400, William Jordan wrote:
> >
> > The udev support/naming works fine. Regarding naming, the ib_uverbs
> > and ib_umad drivers only put port specific devices in the /dev/infiniband
> > directory. They also create non-port specific device nodes 
> > /dev/infiniband_mad
> > and /dev/infiniband_verbs. Would it be more consistent to create a
> > /dev/infiniband_cm instead of /dev/infiniband/cm?
> 
>  You are right, I hadn't noticed the distinction before, looks like
> /dev/infiniband_cm is the way to go.
> 
> -Libor
> 

Actually, I think I'm wrong. On my system, /dev/infiniband_mad is the
same device as /dev/infiniband/umad1, and /dev/infiniband_verbs is the
same device as /dev/infiniband/uverbs0.

Should the device nodes /dev/infiniband_mad and /dev/infiniband_verbs
exist at all? Are they artifacts of the way the classes work, or a
setup problem on my system? Do other people have these nodes?

Unless I've gotten it wrong (again), I believe /dev/infiniband/ucm (as
it exists) is an appropriate device name for the userlevel cm.

Sorry for the confusion.

-- 
Bill Jordan
SilverStorm Technologies
_______________________________________________
openib-general mailing list
[email protected]
http://openib.org/mailman/listinfo/openib-general

To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general

Reply via email to