> > This commit optionally enables a hardware reset of the lan8650/1 > > mac-phy. These chips have a software reset that is discourage from use > > in the manual since it only resets the internal phy. > The software reset done by the current driver is not only resetting the > internal PHY, it resets the entire MAC-PHY including the integrated PHY. > The reset bit of the Clause 22 basic control register only will reset > the internal PHY alone. But oa_tc6_sw_reset_macphy() function is writing > software reset bit in the Reset Control and Status register which resets > the entire MAC-PHY including the internal PHY. All right, I did not dig deep enough obviously. > The above note is given in the lan8650 datasheet to let the user to know > that clause 22 software reset will reset only internal PHY but I don't > think they mean it for the MAC-PHY software reset done from Reset > Control and Status register. Could still be relevant to implement the .soft_reset with -EOPNOTSUPP as Andrew has suggested in the phy driver. > > So in my opinion, I don't see the need of external pin reset as the > existing oa_tc6_sw_reset_macphy() function does the software reset of > the entire MAC-PHY. I agree with your assesment that this invalidates the problem I was aiming at solving. Additionally I figured out why my setup did not work without the HW reset, I had missed a pull resistor in the schematic that held the IC in reset. To me it seems more feature complete to have a driver option for the physical capabilities of the chip, but if it doesn't actually solve a problem it might just be bloat. > > Still if you see a need to have this external pin reset as an optional > function then it may be needed for all the vendor specific MAC drivers. > In that case, reset-gpios parameter value alone can be taken from the > chip specific device tree and the remaining code for operating the reset > gpio can be moved to oa_tc6.c and the function name can be > oa_tc6_hw_reset_macphy(). > If the consensus is to keep a HW reset I do like this suggestion. I won't push for this to be included, if it is, I'm happy to address the feedback of patch 13. R
