[openib-general] [PATCH] Initial checkin of userspace MAD access

[Michael S. Tsirkin]

Hello!
Quoting r. Johannes Erdfelt (johannes at erdfelt.com) "Re: [openib-general] [PATCH] Initial checkin of userspace MAD access":
> On Wed, Nov 03, 2004, Roland Dreier <roland at topspin.com> wrote:
> > By the way, buried down at the end of the patch is some documentation
> > about creating device files:
> > 
> > +/dev files
> > +
> > +  To create the appropriate character device files automatically with
> > +  udev, a rule like
> > +
> > +    KERNEL="umad*", NAME="infiniband/%s{ibdev}/umad%s{port}"
> > +
> > +  can be used.  This will create nodes such as /dev/infiniband/mthca0/umad1
> > +  for port 1 of device mthca0.
> > 
> > Do the names /dev/infiniband/mthca0/umad1 and so on make sense to
> > people?  I thought that userspace verbs support would probably use a
> > file like /dev/infiniband/mthca0/verbs, etc.
> > 
> > In any case, now is probably the time to object before we have legacy
> > issues to worry about....
> 
> Does the device name need to have the HCA driver name in it? Also, the u
> in umad is implied.
> 
> Wouldn't it be more appropriate to do something like this:
> 
> /dev/infiniband/hca0/mad1
> 
> or maybe even:
> 
> /dev/ib/hca0/mad1
> 
> JE

I'd suggest

/dev/ib/hca0/ports/1/mad

Then the user can give the file name like
/dev/ib/hca0/ports/1 to opensm directly, open sm will
just append "/mad", and it is also easier to find out
how many ports in hca0 without switching to sysfs.

MST