Re: [RANT] Linux-IrDA status

From: Jean Tourrilhes (jt@spica.hpl.hp.com)
Date: Wed Nov 08 2000 - 13:20:42 EST


On Wed, Nov 08, 2000 at 12:04:05PM +0000, Dag Brattli wrote:
>
> Hi,
>
> It was implemented this way because the IrDA device drivers are implemented
> like normal network device drivers and didn't want to mess with struct netdevice
> in order to change the speed of the driver. I decided to use ioctl since it had to
> be possible to do this from user-space (sniffers) as well as from the IrDA stack.
> The only thing the IrDA stack knows about is the netdevice. Some frames we
> receive will trigger a speed change which we must handle from within the
> stack (so it's inside the bh and not actually in the "hard" interrupt)
>
> What do you suggest I do?
>
> 1. Add a change_speed() function to struct netdevice
> 2. Add a protocol specific pointer to struct netdevice
> 3. Embed the speed in skb->cb and send down empty frames
> when I want to change the speed without transmitting anything.
> 4. Anything else?

        If somebody tell us which is the "right way", I'll try to code
that ASAP. We take any feedback very seriously ;-)
        I personally would go with #1, because "struct netdevice" is
full of protocol specific stuff anyway... And if we do our job right,
it can be reused for other stuff.

        Any comments ?

> -- Dag

        Jean
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Wed Nov 15 2000 - 21:00:12 EST