On Sun, Dec 09, 2001 at 09:56:38PM +1100, Brad Hards wrote:

> After an abortive previous attempt at providing user space reporting
> (http://marc.theaimsgroup.com/?t=100677748400003&r=1&w=2), I implemented the
> ioctl() support for the ethtool user space tool (http://gkernel.sf.net) in the
> catc driver. Here is an example:
> [bradh@leon linux]# ethtool eth1
> Settings for eth1:
>       Link detected: no
> [bradh@leon linux]# ethtool eth1
> Settings for eth1:
>       Link detected: yes
> [bradh@leon linux]# ethtool -i eth1
> driver: EL1210A NetMate USB Ethernet
> version: v2.8
> firmware-version: 
> bus-info: usb1:9

Nice. Send it to Greg KH, when you want to have included it in the
kernel, it's OK with me.

> Patch is attached, against 2.4.17-pre6.
> 
> Discussion issue:
> The CATC device provides link state as interrupt status bits. So the link
> state reporting is only accurate when interrupts are being delivered. The catc
> driver currently only submits the interrupt URB in its open() routine. All
> this means is that it only works if the interface is UP.
> Should we try for a sensible handling of link state (by submitting the
> interrupt URB in probe()), even if it means possibly handling junk interupts? 

I think it is fairly OK, the only problem is that it'll eat a slice out
of the available USB bandwith. The interrupt is polled in every USB
frame.

-- 
Vojtech Pavlik
SuSE Labs

_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to