Re: [PATCH] arm64: dts: rockchip: add EEPROM node for NanoPi R4S

From: Chen-Yu Tsai
Date: Mon Jun 07 2021 - 13:08:50 EST


On Tue, Jun 8, 2021 at 1:04 AM Johan Jonker <jbx6244@xxxxxxxxx> wrote:
>
>
>
> On 6/7/21 6:01 PM, Tianling Shen wrote:
> > Hi Johan,
> >
> > On Mon, Jun 7, 2021 at 6:26 PM Johan Jonker <jbx6244@xxxxxxxxx> wrote:
> >>
> >> Hi Chen-Yu,
> >>
> >> On 6/7/21 11:40 AM, Chen-Yu Tsai wrote:
> >>> On Mon, Jun 7, 2021 at 5:31 PM Johan Jonker <jbx6244@xxxxxxxxx> wrote:
> >>>>
> >>>> Hi Tianling,
> >>>>
> >>>> On 6/7/21 10:17 AM, Tianling Shen wrote:
> >>>>> NanoPi R4S has a EEPROM attached to the 2nd I2C bus (U92), which
> >>>>> stores the MAC address.
> >>>>>
> >>>>> Signed-off-by: Tianling Shen <cnsztl@xxxxxxxxx>
> >>>>> ---
> >>>>> arch/arm64/boot/dts/rockchip/rk3399-nanopi-r4s.dts | 9 +++++++++
> >>>>> 1 file changed, 9 insertions(+)
> >>>>>
> >>>>> diff --git a/arch/arm64/boot/dts/rockchip/rk3399-nanopi-r4s.dts b/arch/arm64/boot/dts/rockchip/rk3399-nanopi-r4s.dts
> >>>>> index cef4d18b599d..4a82f50a07c5 100644
> >>>>> --- a/arch/arm64/boot/dts/rockchip/rk3399-nanopi-r4s.dts
> >>>>> +++ b/arch/arm64/boot/dts/rockchip/rk3399-nanopi-r4s.dts
> >>>>> @@ -68,6 +68,15 @@
> >>>>> status = "disabled";
> >>>>> };
> >>>>>
> >>>>> +&i2c2 {
> >>>>> + eeprom@51 {
> >>>>> + compatible = "microchip,24c02", "atmel,24c02";
> >>>>> + reg = <0x51>;
> >>>>> + pagesize = <16>;
> >>>>
> >>>>> + read-only; /* This holds our MAC */
> >>>>
> >>>> The mainline dts files should be generic I think.
> >>>> Any comment about "use", partitions or write ability should be avoided.
> >>>> It's up the user.
> >>>
> >>
> >>> Per the datasheet for this specific EEPROM, the latter half (128 bytes)
> >>> is read-only in hardware by design though.
> >>
> >> The 24AA02XEXX is programmed at the factory with a
> >> globally unique node address stored in the upper half
> >> of the array and permanently write-protected. The
> >> remaining 1,024 bits are available for application use.
> >>
> >
>
> > In my opinion, as this contains data programmed by the factory, would
> > it be okay to keep it read-only?
>
> This chip is not completely read-only.
> There might be users that like to try some other mac_address or store
> something else in that lower part. Is this then still possible?
> Generic DT describes hardware independent from what Linux drivers or
> other OS are capable off.
> This factory mac_addres is permanently write-protected, so no need to
> keep the rest read-only.
>
> nvmem-cells = <&new_mac_address_in_lower_part>;
> nvmem-cells-names = "mac-address";
>
> >
> >> Just a question...
> >>
> >> nvmem-cells = <&mac_address>;
> >> nvmem-cells-names = "mac-address";
> >>
> >> Which part does this point to?
> >>
> >> Can we use the lower part to store/rewrite this too?
> >>
> >> ===
> >>
> >> From at24.yaml:
> >>
> >> items:
> >> - pattern:
> >> "^(atmel|catalyst|microchip|nxp|ramtron|renesas|rohm|st),(24(c|cs|lc|mac)[0-9]+|spd)$"
> >> - pattern: "^atmel,(24(c|cs|mac)[0-9]+|spd)$"
> >>
> >> How does Microchip 24AA025E48 fit the regex?
> >> What compatible would you advise?
> >
> > It seems that 24AA025E48 is a variant of 24MAC402 [1], and
> > `atmel,24c02` will be okay in this case.
> > 1. https://lkml.org/lkml/2018/1/24/494
>
> Ask Heiko. ;)
>
> As long as it does not generate more notifications then we already have.

I think having a part specific compatible would be better. That way if
someone wanted to implement read-only "feedback" to users for the second
half they could.

ChenYu

> >
> > Thanks,
> > Tianling.
> >
> >>
> >> ===
> >>
> >> Johan
> >>
> >>>
> >>> ChenYu
> >>>