Re: [PATCH ethtool 2/2] Ethtool: Implements PHY Loopback

On 28/11/16 09:56, Florian Fainelli wrote:

On 11/28/2016 05:25 AM, Allan W. Nielsen wrote:

quoted

From: Raju Lakkaraju <redacted>

Add loopback in ethtool tunables to access PHY drivers.

Ethtool Help: ethtool -h for PHY tunables
    ethtool --set-phy-tunable DEVNAME      Set PHY tunable
                [ loopback off|near|far|extn ]
    ethtool --get-phy-tunable DEVNAME      Get PHY tunable
                [ loopback ]

Ethtool ex:
  ethtool --set-phy-tunable eth0 loopback near
  ethtool --set-phy-tunable eth0 loopback far
  ethtool --set-phy-tunable eth0 loopback extn
  ethtool --set-phy-tunable eth0 loopback off

  ethtool --get-phy-tunable eth0 loopback

Signed-off-by: Raju Lakkaraju <redacted>
Signed-off-by: Allan W. Nielsen <redacted>
---

quoted

+Near-End Loopback:
+Transmitted data (TXD) is looped back in the PCS block onto the receive data
+signal (RXD). When Near-End loopback enable, no data is transmitted over
+the network. no data receive from the network.

This is also known as the local loopback test mode, right?

Yes - Traffic transmitted/generated by the host list loopback to the
host instead of transmitting it on the wire.

quoted

+
+Far-End Loopback:
+This loopback is a special test mode to allow testing the PHY from link
+partner side. In this mode data that is received from the link partner pass
+through the PHY's receiver, looped back on the MII and transmitted back to
+the link partner.

And this is the remote loopback mode.

Yes - Traffic receiwed on the "wire" is transmitted back on the wire.

quoted

+
+External Loopback:
+An RJ45 loopback cable can be used to route the transmit signals an the
+output of the trnsformer back to the receiver inputs and this loopback will
+work at either 10 or 100 or 1000 Mbps speed.
+RJ45 Loopback cable created by conncting pin 1 to pin 3 and connecting pin
+2 to pin 6.

OK, this name makes sense to me, but for the two other names, we need to
use a terminology that is clearer to the reader and/or people familiar
and targeted at using this feature (e.g: in a lab or during manufacturing).

Sure, we can find better names and/or improve the documentation. But
before jumping to that, then it is properly a good idea to agree on
the overall concept.

We will get back to the naming when we agree on the other parts.

/Allan