Re: Flaw in the driver-model implementation of attributes

[Greg KH <greg@kroah.com>]

On Thu, Jun 19, 2003 at 05:14:42PM -0400, Alan Stern wrote:
> 
> So what Mike says the SCSI core does, and what Pat says it ought to do, 
> isn't what actually happens.  Is this simply an indication that the work 
> of completing the SCSI code to the driver model isn't yet complete?  It 
> seems as though a very large change would be needed to make things work as 
> Pat described.

I think it's more of the matter that Mike is still changing the way the
SCSI core works with the class and driver model.  I suggest you take
this to the linux-scsi list to hash out the way scsi should and does
work in order to find the best solution for usb-storage, as I guess that
it also would be the best solution for all other SCSI host drivers.

> In the meantime, where should I register my class device for the 
> usb-storage driver?

Why not hold off until the scsi people are finished with their changes?
If after that, you _really_ need to be putting usb-storage only
attributes in the sysfs tree somewhere, we can talk again.

thanks,

greg "scsi makes my head hurt" k-h