On Mon, Apr 05, 2021 at 09:07:34AM +0000, Voon, Weifeng wrote: > > On Fri, Apr 02, 2021 at 07:45:04AM +0000, Voon, Weifeng wrote: > > > > > + /* 2.5G mode only support 2500baseT full duplex only */ > > > > > + if (priv->plat->has_gmac4 && priv->plat->speed_2500_en) { > > > > > + phylink_set(mac_supported, 2500baseT_Full); > > > > > + phylink_set(mask, 10baseT_Half); > > > > > + phylink_set(mask, 10baseT_Full); > > > > > + phylink_set(mask, 100baseT_Half); > > > > > + phylink_set(mask, 100baseT_Full); > > > > > + phylink_set(mask, 1000baseT_Half); > > > > > + phylink_set(mask, 1000baseT_Full); > > > > > + phylink_set(mask, 1000baseKX_Full); > > > > > > > > Why? This seems at odds to the comment above? > > > > > > > What about 2500baseX_Full ? > > > > > > The comments explain that the PCS<->PHY link is in 2500BASE-X and why > > > 10/100/1000 link speed is mutually exclusive with 2500. > > > But the connected external PHY are twisted pair cable which only > > > supports 2500baseT_full. > > > > The PHY should indicate what modes its supports. The PHY drivers > > get_features() call should set supported to only 2500baseT_Full, if that is > > all it supports. > > > > What modes are actually used should then be the intersect of what both the > > MAC and the PHY indicate they can do. > > Noted Andrew. Instead of masking the 10/100/1000 mode support in the MAC, we > will > set the supported modes in the PCS.
PCS? You said: > > > But the connected external PHY are twisted pair cable which only > > > supports 2500baseT_full. So it should be the PHY, not the PCS, which indicates it only supports 2500baseT_full. Andrew