[quagga-dev 3628] Re: IFF_RUNNING and link-detect

Paul Jakma paul at clubi.ie
Tue Sep 6 15:58:23 BST 2005

On Tue, 6 Sep 2005, Jens Schweikhardt wrote:

> # Note that at least Quagga maintainer would "violently" disagree with
> # this.
> That would be you?

Ah no, I only agree with maintainer in principle. :)

> I've heard that people think the link status does not fit the IETF 
> model of only AdminStatus and OperStatus, but rather should be 
> treated as a third status. My very limited understanding is that 
> making IFF_RUNNING represent the link status would be making the 
> link status determine the OperStatus...

IFF_RUNNING has essentially lost any meaning it had before (I think 
once upon a time it meant "resources are allocated"). So on Linux at 
least, and Solaris I think, it does indeed directly reflect the 
operational state - if that is directly equivalent to link-state.

> # You should implement whatever link-state functionality
> # is specified for network drivers to implement (eg something to do
> # with "netif_carrier" for linux drivers).
> Thanks for the pointer; will look into it.

Following functions in right places should do it:

 	netif_carrier_{on,off} (struct net_device *)

The Linux 'netdev' list on vger.kernel.org is best place to get 
advice on linux network driver issues.

Paul Jakma	paul at clubi.ie	paul at jakma.org	Key ID: 64A2FF6A
Don't get mad, get interest.

More information about the Quagga-dev mailing list