Re: [PATCH] smsc95xx: support ethtool get_regs

["Émeric Vigier" <emeric.vigier@savoirfairelinux.com>]

----- Mail original -----
> De: "Francois Romieu" <romieu@fr.zoreil.com>
> À: "Émeric Vigier" <emeric.vigier@savoirfairelinux.com>
> Cc: "Steve Glendinning" <steve@shawell.net>, "steve glendinning" <steve.glendinning@smsc.com>,
> netdev@vger.kernel.org, "Nancy Lin" <nancy.lin@smsc.com>
> Envoyé: Vendredi 6 Juillet 2012 16:01:46
> Objet: Re: [PATCH] smsc95xx: support ethtool get_regs
> 
> Émeric Vigier <emeric.vigier@savoirfairelinux.com> :
> [...]
> > +static int smsc95xx_ethtool_getregslen(struct net_device *dev)
> > +{
> > +	/* all smsc95xx registers plus all phy registers */
> > +	return COE_CR - ID_REV + 1 + 32 * sizeof(u32);
> 
> I do not see where ID_REV is accounted for in the loops below.
> 
> s/32 */PHY_SPECIAL */ or s/PHY_SPECIAL/32/ below.

I will go for the second proposal. I love your sed syntax btw :-)

> 
> I thought PHY registers were 16 bits wide. Moreover they are already
> available through smsc95xx_ioctl().

Yes, there are 16 bits wide according to smsc95xx.h.
But other smsc drivers define 32bit wide PHY regs. I made myself believe that smsc would use the same PHY for each ethernet chip.
So would something like s/32 * sizeof(u32)/PHY_SPECIAL * sizeof(u16)/ solve the issue here?

Concerning the ioctl, I found ethtool much easier to use. And I believe smsc9514 is a very popular chipset, so this could help others debugging it.

> 
> > +}
> > +
> > +static void
> > +smsc95xx_ethtool_getregs(struct net_device *netdev, struct
> > ethtool_regs *regs,
> > +			 void *buf)
> > +{
> > +	struct usbnet *dev = netdev_priv(netdev);
> > +	unsigned int i, j = 0, retval;
> 
> 	unsigned int i, j, retval;
> 
> > +	u32 *data = buf;
> > +
> > +	netif_dbg(dev, hw, dev->net, "ethtool_getregs\n");
> 
> The tracing framework does provide almost the same information.

Do you mean LTT? I am not familiar with it, I should have a look.
I remove netif_dbg then.

> 
> > +
> > +	retval = smsc95xx_read_reg(dev, ID_REV, &regs->version);
> > +	if (retval < 0) {
> > +		netdev_warn(dev->net, "REGS: cannot read ID_REV\n");
> 
> s/dev->net/netdev/ ?

You are right, I also changed smsc95xx_ethtool_getregslen() definition to match this syntax.

> 
> 
> > +		return;
> > +	}
> > +
> > +	for (i = 0; i <= COE_CR; i += (sizeof(u32))) {
> > +		retval = smsc95xx_read_reg(dev, i, &data[j++]);
> 
> 	for (i = 0, j = 0; i <= COE_CR; i += sizeof(u32), j++) {
> 		retval = smsc95xx_read_reg(dev, i, data + j);

I should change that in previous "for" loop as well I suppose?

> 
> --
> Ueimor
> 

Emeric