Re: [PATCH] net: add QCA alx Ethernet driver

[Joe Perches <joe@perches.com>]

On Fri, 2012-03-02 at 00:40 +0100, Francois Romieu wrote:
> Stephen Hemminger <shemminger@vyatta.com> :
> [...]
> > Evolution is better. The new driver has lots of new callbacks to handle
> > the fact it is dealing with two different chipsets. Not only that your callbacks
> > are built at runtime which leads to security concerns.
> > 
> > There is a reason the two Marvell based drivers (skge and sky2) are
> > different drivers. Having to do extra per-chip callbacks is a clear sign
> > the driver should be split in two.
> 
> (arguable)
[]
> > +void alx_hw_printk(const char *level, const struct alx_hw *hw,
> > +		   const char *fmt, ...)
> > +{
> > +	struct va_format vaf;
> > +	va_list args;
> > +
> > +	va_start(args, fmt);
> > +	vaf.fmt = fmt;
> > +	vaf.va = &args;
> > +
> > +	if (hw && hw->adpt && hw->adpt->netdev)
> > +		__netdev_printk(level, hw->adpt->netdev, &vaf);
> > +	else
> > +		printk("%salx_hw: %pV", level, &vaf);
> > +
> > +	va_end(args);
> > +}
> 
> Designing a new logging facility smells like being unable to figure
> the current context.
> 
> And the printk does not even use KERN_... :o(

Sure it does.

level is the first argument and is emitted as %s

> > +/*
> > + *  alx_validate_mac_addr - Validate MAC address
> > + */
> > +static int alx_validate_mac_addr(u8 *mac_addr)
> > +{
> > +	int retval = 0;
> > +
> > +	if (mac_addr[0] & 0x01) {
> > +		printk(KERN_DEBUG "MAC address is multicast\n");
> > +		retval = -EADDRNOTAVAIL;
> > +	} else if (mac_addr[0] == 0xff && mac_addr[1] == 0xff) {
> > +		printk(KERN_DEBUG "MAC address is broadcast\n");
> > +		retval = -EADDRNOTAVAIL;
> > +	} else if (mac_addr[0] == 0 && mac_addr[1] == 0 &&
> > +		   mac_addr[2] == 0 && mac_addr[3] == 0 &&
> > +		   mac_addr[4] == 0 && mac_addr[5] == 0) {
> > +		printk(KERN_DEBUG "MAC address is all zeros\n");
> > +		retval = -EADDRNOTAVAIL;
> > +	}
> > +	return retval;
> > +}
> 
> Bloat. It should use is_valid_ether_addr.

I sent patches already.
It's also out of order, testing multicast then broadcast
instead of the reverse.