> -----Original Message----- > From: Andrew Lunn <andrew@xxxxxxx> > Sent: 2022年8月18日 9:44 > To: Wei Fang <wei.fang@xxxxxxx> > Cc: hkallweit1@xxxxxxxxx; linux@xxxxxxxxxxxxxxx; davem@xxxxxxxxxxxxx; > edumazet@xxxxxxxxxx; kuba@xxxxxxxxxx; pabeni@xxxxxxxxxx; > robh+dt@xxxxxxxxxx; krzysztof.kozlowski+dt@xxxxxxxxxx; f.fainelli@xxxxxxxxx; > netdev@xxxxxxxxxxxxxxx; devicetree@xxxxxxxxxxxxxxx; > linux-kernel@xxxxxxxxxxxxxxx > Subject: Re: [PATCH net 1/2] dt: ar803x: Document disable-hibernation > property > > > Yes, after the PHY enters hibernation mode that the RX_CLK stop > > ticking, but for stmmac, it is essential that RX_CLK of PHY is present > > for software reset completion. Otherwise, the stmmac is failed to > > complete the software reset and can not init DMA. > > So the RX_CLK is more than the recovered clock from the bit stream on the > wire. The PHY has a way to generate a clock when there is no bit stream? > Yes, when disable hibernation mode, the RX_CLK always output a valid clock. > To me, it sounds like your hardware design is wrong, and it should be using the > 25MHz reference clock. And what you are proposing is a workaround for this > hardware problem. > > Anyway, i agree with Russell, a DT property is fine. But please make it clear in > the binding documentation that disabling hibernation has the side affect of > keeping the RX_CLK ticking when there is no link. That is probably what people > want this for, not to actual disable hibernation. > > Andrew
