[U-Boot] [PATCH V2 4/5] net: fec: sharing MDIO for two enet controllers

[Joe Hershberger]

On Wed, Mar 21, 2018 at 1:59 PM, Jagan Teki <jagannadh.teki at gmail.com> wrote:
> On Thu, Mar 22, 2018 at 12:09 AM, Joe Hershberger
> <joe.hershberger at ni.com> wrote:
>> On Wed, Mar 21, 2018 at 1:29 PM, Jagan Teki <jagannadh.teki at gmail.com> wrote:
>>> On Wed, Mar 21, 2018 at 6:54 PM, Peng Fan <peng.fan at nxp.com> wrote:
>>>>> >> >
>>>>> >> >
>>>>> >> >> -----Original Message-----
>>>>> >> >> From: Jagan Teki [mailto:jagannadh.teki at gmail.com]
>>>>> >> >> Sent: 2018年3月21日 17:19
>>>>> >> >> To: Peng Fan <peng.fan at nxp.com>
>>>>> >> >> Cc: Joe Hershberger <joe.hershberger at ni.com>; Fabio Estevam
>>>>> >> >> <fabio.estevam at nxp.com>; U-Boot Mailing List
>>>>> >> >> <u-boot at lists.denx.de>
>>>>> >> >> Subject: Re: [U-Boot] [PATCH V2 4/5] net: fec: sharing MDIO for
>>>>> >> >> two enet controllers
>>>>> >> >>
>>>>> >> >> On Wed, Mar 21, 2018 at 2:31 PM, Peng Fan <peng.fan at nxp.com>
>>>>> wrote:
>>>>> >> >> > On i.MX6SX, 6UL and 7D, there are two enet controllers each has
>>>>> >> >> > a MDIO port. But Some boards share one MDIO port for the two enets.
>>>>> >> >> > So introduce a configuration CONFIG_FEC_MXC_MDIO_BASE to
>>>>> >> >> > indicate the MDIO port for sharing.
>>>>> >> >> >
>>>>> >> >> > To i.MX28, adapt to use the new config
>>>>> >> >> >
>>>>> >> >> > Signed-off-by: Peng Fan <peng.fan at nxp.com>
>>>>> >> >> > Acked-by: Joe Hershberger <joe.hershberger at ni.com>
>>>>> >> >> > Cc: Fabio Estevam <fabio.estevam at nxp.com>
>>>>> >> >> > ---
>>>>> >> >> >
>>>>> >> >> > V2:
>>>>> >> >> >  adapt mx28 to use FEC_MXC_MDIO_BASE
>>>>> >> >> >
>>>>> >> >> >  drivers/net/Kconfig       | 9 ++++++++-
>>>>> >> >> >  drivers/net/fec_mxc.c     | 8 ++++++--
>>>>> >> >> >  include/configs/mx28evk.h | 1 +
>>>>> >> >> >  3 files changed, 15 insertions(+), 3 deletions(-)
>>>>> >> >> >
>>>>> >> >> > diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig index
>>>>> >> >> > de1947ccc1..c5fd6f648a 100644
>>>>> >> >> > --- a/drivers/net/Kconfig
>>>>> >> >> > +++ b/drivers/net/Kconfig
>>>>> >> >> > @@ -147,9 +147,16 @@ config ETHOC
>>>>> >> >> >         help
>>>>> >> >> >           This MAC is present in OpenRISC and Xtensa XTFPGA
>>>>> boards.
>>>>> >> >> >
>>>>> >> >> > +config FEC_MXC_MDIO_BASE
>>>>> >> >> > +       hex "MDIO base address for the FEC controller"
>>>>> >> >> > +       depends on FEC_MXC
>>>>> >> >> > +       help
>>>>> >> >> > +         This specifies the MDIO registers base address. It is used
>>>>> when
>>>>> >> >> > +         two FEC controllers share MDIO bus.
>>>>> >> >>
>>>>> >> >> How about retrieving this from dt? we have board that support dt
>>>>> >> >> along with dm supporting.
>>>>> >> >
>>>>> >> > It's ok to retrieve the info from dt, but we are not only support
>>>>> >> > dt, we also need to support non-dt case. Define FEC_MXC_MDIO_BASE
>>>>> >> > is the simplest method to achieve that I think.
>>>>> >>
>>>>> >> But patch adds the same for DM case which usually retrieve the info
>>>>> >> from dt, point here is to get rid of ifdef and new CONFIG_ ie where
>>>>> >> DM and dts play smart atleast.
>>>>> >
>>>>> > The uboot fec_mxc driver or net driver is not that sync with linux
>>>>> > code, Using dt here, need to parse phy-handle = <&ethphy0>;, then
>>>>> > parse ethphy0 parent to get reg, I just think this like hack.
>>>>>
>>>>> No, I think we can do this with adding new DM MDIO similar to DM PHY which
>>>>> recently done. May be some sort of efforts but it is permanent.
>>>> We do not have that driver now, so could we first have this patch? When
>>>> DM MDIO ready, this piece code could be removed then?
>>>
>>> ie. up to Joe. Honestly this macro become removed in future, my point
>>> here is why we need to maintain dead macro instead of adding proper
>>> maintainable stuff. I'm pretty sure adding DM_MDIO is straight forward
>>> and as of now just add what we need and rest will implement future.
>>> You may become victim to others to move DM_ETH as soon as possible :)
>>
>> It would be ideal if you wanted to implement DM MDIO, but I can also
>> appreciate that this is not already there for you to use. As I
>> commented when I acked this, the I'm OK with this approach at this
>> time due to the state of the DM support in eth.
>
> As I said before, its your decision.
>
>>
>> As a side note, maybe moving other boards that use this NIC to DM_ETH
>> and removing non-DM support would be a better cleanup to start with.
>
> To be honest this can't be predictable (until you make a note on
> MIGRATION deadline), I still see new boards from i.MX still can't use
> dt.

That's not good. Simon and I (and I thought everyone else) have been
operating under the declaration that no new driver or board can be
added using legacy. I hope whoever is acking this board support will
stop.

-Joe