Re: [PATCH -mm 5/9] netconsole: Introduce dev_status member

[Stephen Hemminger <shemminger@linux-foundation.org>]

On Wed, 4 Jul 2007 08:56:42 -0500
Matt Mackall <mpm@selenic.com> wrote:

> On Wed, Jul 04, 2007 at 04:38:04PM +0530, Satyam Sharma wrote:
> > From: Satyam Sharma <ssatyam@cse.iitk.ac.in>
> > 
> > [5/9] netconsole: Introduce dev_status member
> > 
> > Introduce a new member in netconsole_target that tracks the status (up or
> > down) of the underlying interface network device that the specific logging
> > target netpoll is attached to.
> > 
> > We then join this up with the just-introduced net_device notifier, and
> > introduce NETDEV_UP and NETDEV_DOWN notifications. By disabling the target
> > when the corresponding local interface is down, we save the overhead of
> > unnecessarily disabling interrupts and calling into the netpoll stack in
> > console->write().
> 
> Yuck. 
> 
> > +/*
> > + * Why no net_dev_is_up() in netdevice.h? The kernel could lose a lot of
> > + * weight if only netdevice.h had the good sense to export such a function.
> > + * Oh well ...
> > + */
> > +static inline int net_dev_is_up(struct net_device *net_dev)
> > +{
> > +	return ((net_dev->flags & IFF_UP) == IFF_UP);
> > +}
> 
> Why editorialize? Why not just add this to netdevice.h?
> 
> > +	if (nt->dev_status) {
> 
> Why not simply call net_dev_is_up?

The flags field values are really BSD legacy-ish stuff and should not be
used internally. IFF_UP etc, do not have the necessary atomic properties.

Use netif_running() instead.