On Fri, May 05, 2000, Alan Cox <[EMAIL PROTECTED]> wrote:
> > The point being that the implement the device/enpoint scheme, you need a
> > dynamic filesystem. Assigning fixed major/minors for device/endpoints is
> 
> No I dont
> 
> > unfortunately not possible as it would exhaust the available major/minor
> > number space accounting for all possible combinations.
> 
> Agreed
> 
> 
>       struct usb_select
>       {
>               int bus;
>               int device;
>               int ep;
>       };
> 
>       ioctl(fd, USBIOCSELECT, &usb_select);
> 
> >From that point the fd is bound to that given bus,device,ep - its a way to
> work around the device space problem. Its not pretty but it works.

Exactly, you are making compromises for deficiencies in the system. 

Since you can't make it a part of the namespace, you move it to the ioctl
space.

This is a step backwards.

With a dynamic filesystem, this isn't a problem.

JE


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to