On Thursday 15 June 2006 1:36 pm, Greg KH wrote:
> On Thu, Jun 15, 2006 at 01:07:32PM -0700, David Brownell wrote:
> > On Wednesday 14 June 2006 11:33 pm, Greg KH wrote:
> >
> > > + snprintf(ep_dev->dev.bus_id, BUS_ID_SIZE, "usbdev%d.%d_ep%02x",
> > > + udev->bus->busnum, udev->devnum,
> > > + endpoint->desc.bEndpointAddress);
> > > +
> > > + retval = device_register(&ep_dev->dev);
> > > + if (retval)
> > > + goto error;
> > > + sysfs_create_group(&ep_dev->dev.kobj, &ep_dev_attr_grp);
> > > +
> > > + endpoint->ep_dev = ep_dev;
> > > +
> > > + /* create the symlink to the old-style "ep_XX" directory */
> > > + sprintf(name, "ep_%02x", endpoint->desc.bEndpointAddress);
> > > + sysfs_create_link(&parent->kobj, &endpoint->ep_dev->dev.kobj, name);
> >
> > Hmm, would anyone object if this changed to use the epNin/epNout
> > convention instead of ep8N/ep0N? (Better yet, using decimal not
> > hex ...) That's the convention used everywhere else that we
> > print an endpoint descriptions, since it less obfuscatory. :)
>
> Hm, should I just change the usbdevX.X_epXX names to also use the
> epXin/epXout suffix? I like it a bit better too, otherwise I always
> have to go look up the bit for in/out :)
Yes, I think everything should use the "epN{in,out}" convention.
Not many of us are sick enough to know, even at 7am before coffee
enters the bloodstream, what bit 7 indicates. ;)
> Hm, but would that really work ok for usbfs2? The system already knows
> the endpoint numbers from the config space, and should not have to do a
> translation on it to find the proper device node, right?
It should work easily for usbfs2, yes. It's just a more useful way
to present the same information.
- Dave
_______________________________________________
linux-usb-devel@lists.sourceforge.net
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
