Hi Florian, Thank you for review comments. On Mon, Oct 17, 2016 at 05:38:46AM -0700, Florian Fainelli wrote: > EXTERNAL EMAIL > > > On October 17, 2016 12:31:54 AM PDT, Raju Lakkaraju <Raju.Lakkaraju@xxxxxxxxxxxxx> wrote: > >Hi Andrew, > > > >Thank you for code review and comments. > > > >On Fri, Oct 14, 2016 at 02:12:32PM +0200, Andrew Lunn wrote: > >> EXTERNAL EMAIL > >> > >> > >> On Fri, Oct 14, 2016 at 05:10:32PM +0530, Raju Lakkaraju wrote: > >> > From: Raju Lakkaraju <Raju.Lakkaraju@xxxxxxxxxxxxx> > >> > > >> > For operation in cabling environments that are incompatible with > >> > 1000BAST-T, VSC8531 device provides an automatic link speed > >> > downshift operation. When enabled, the device automatically changes > >> > its 1000BAST-T auto-negotiation to the next slower speed after > >> > a configured number of failed attempts at 1000BAST-T. > >> > This feature is useful in setting up in networks using older cable > >> > installations that include only pairs A and B, and not pairs C and > >D. > >> > >> Any reason not to just turn this on by default when auto-neg is > >> enabled? > >> > >Downshift can enable by default when auto-neg enabled. This is good > >idea. > >But we would like to provide option to customer can choose whether this > >feature need to enable or disable and also configure failure attempts. > > > >Do you have any other suggestion how to configure failure attempts? > > Is the speed downshift feature similar to what Intel and Broadcom refer to as wirespeed? I have seen cases with Broadcom PHYs where we had to turn such a feature on to allow auto-negotiation to complete with 4-wire cables, but this had the downside of impacting normal autoneg, so it is left disabled. > Yes. I check the Broadcom wirespeed code. Downshift is similar to wirespeed. But Broadcom wirespeed configuration in Ethernet controller. > I would expect the number of customers using this feature to be fairly limited, so having a tunable to turn this downshift on/off may be acceptable. Ethtool supports a number of tunable parameters now (such as rx_copybreak), there may be room for using something similar for boolean flags like these. > This implementation shows little bit specific to Ethernet controller. Do you have any PHY specific examples? In another mail thread, you proposed similar to net device features. Shall i implement that suggestion here? > -- > Florian --- Thanks, Raju. -- To unsubscribe from this list: send the line "unsubscribe devicetree" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
