netdev
[Top] [All Lists]

Re: [PATCH] dummy -- support hardware address

To: "David S. Miller" <davem@xxxxxxxxxx>
Subject: Re: [PATCH] dummy -- support hardware address
From: Stephen Hemminger <shemminger@xxxxxxxx>
Date: Thu, 8 Apr 2004 15:09:01 -0700
Cc: netdev@xxxxxxxxxxx
In-reply-to: <20040408145403.34382c96.davem@xxxxxxxxxx>
Organization: Open Source Development Lab
References: <20040408110834.2fd1bd25@xxxxxxxxxxxxxxxxxxxxx> <20040408145403.34382c96.davem@xxxxxxxxxx>
Sender: netdev-bounce@xxxxxxxxxxx
On Thu, 8 Apr 2004 14:54:03 -0700
"David S. Miller" <davem@xxxxxxxxxx> wrote:

> 
> Stephen, why don't you do what usbnet does?  Look at usbnet_init()
> where it goes:
> 
>       get_random_bytes (node_id, sizeof node_id);
>       node_id [0] &= 0xfe;    // clear multicast bit
>       node_id [0] |= 0x02;    // set local assignment bit (IEEE802)
> 
> I distinctly remember Alan Cox saying this was the right way
> do this and avoid conflicts with Vendor assigned IDs.

That would work better, just worried about draining the entropy pool
when I create 1000 dummies in a test script.

<Prev in Thread] Current Thread [Next in Thread>