Re: [PATCH 0/3] phy: qcom: sgmii-eth: add Nord USXGMII support

[Zhangfei Gao]

On Wed, Sep 30, 2026 at 1:16 PM Krzysztof Kozlowski <krzk@xxxxxxxxxx> wrote:
>
> On 29/09/2026 12:49, Zhangfei Gao wrote:
> > Hi, Krzysztof
> >
> > On Mon, Sep 28, 2026 at 6:19 PM Krzysztof Kozlowski <krzk@xxxxxxxxxx> wrote:
> >>
> >> On Wed, Sep 23, 2026 at 04:58:03PM +0800, Zhangfei Gao wrote:
> >>> On Wed, Sep 23, 2026 at 8:04 AM Mohd Ayaan Anwar
> >>> <mohd.anwar@xxxxxxxxxxxxxxxx> wrote:
> >>>>
> >>>> This series adds support for the Ethernet SerDes found on the Qualcomm
> >>>> Nord SoC.
> >>>>
> >>>> Nord uses a QMP v7 SerDes block, which is an updated version of the
> >>>> SGMII SerDes PHY IP already handled by phy-qcom-sgmii-eth. The driver
> >>>> currently supports the 1.25 Gbaud and 3.125 Gbaud configurations used by
> >>>> SGMII, 1000BASE-X and 2500BASE-X. Nord additionally requires the SerDes
> >>>> to operate at 10.3125 Gbaud for its 10 Gigabit Ethernet interfaces.
> >>>>
> >>>> The series can be broadly divided into three parts:
> >>>>
> >>>>   1. Add the qcom,nord-dwmac-sgmii-phy compatible to the binding.
> >>>>   2. Prepare the driver for PHY modes which require different SerDes
> >>>>      programming and readiness checks. Existing SA8775P behaviour is
> >>>>      unchanged.
> >>>>   3. Add the QMP v7 register definitions, 10.3125 Gbaud programming
> >>>>      sequence and readiness checks required for USXGMII and 10GBASE-R.
> >>>>
> >>>> The changes have been tested on the Nord Ride SX platform, which has a
> >>>> MV-Q6223 switch, running in fixed link USXGMII 10G mode.
> >>>>
> >>>> Signed-off-by: Mohd Ayaan Anwar <mohd.anwar@xxxxxxxxxxxxxxxx>
> >>>> ---
> >>>> Mohd Ayaan Anwar (3):
> >>>>       dt-bindings: phy: qcom,sa8775p-dwmac-sgmii-phy: add Nord compatible
> >>>>       phy: qcom: sgmii-eth: prepare for additional PHY modes
> >>>>       phy: qcom: sgmii-eth: add Nord USXGMII support
> >>>>
> >>>>  .../bindings/phy/qcom,sa8775p-dwmac-sgmii-phy.yaml |   1 +
> >>>>  drivers/phy/qualcomm/phy-qcom-qmp-pcs-sgmii.h      |   5 +
> >>>>  drivers/phy/qualcomm/phy-qcom-qmp-qserdes-com-v7.h |   2 +
> >>>>  .../phy/qualcomm/phy-qcom-qmp-qserdes-txrx-v7.h    |   9 +
> >>>>  drivers/phy/qualcomm/phy-qcom-sgmii-eth.c          | 226 ++++++++++++++++++---
> >>>>  5 files changed, 219 insertions(+), 24 deletions(-)
> >>>> ---
> >>>> base-commit: a8c591ed6b672915e0be57843f943a2a723aff40
> >>>> change-id: 20260923-nord_ethernet_serdes-c486f715f383
> >>>>
> >>>> Best regards,
> >>>> --
> >>>> Mohd Ayaan Anwar <mohd.anwar@xxxxxxxxxxxxxxxx>
> >>>>
> >>>
> >>> Good Job，
> >>>
> >>> Tested-by: Zhangfei Gao <zhangfei.gao@xxxxxxxxxxxxxxxx>
> >>
> >> This has to stop.
> >>
> >> Why again adding fake testing to the bindings? You cannot reply twice?
> > Reply twice since I thought they are two different patch set, one is phy,
>
> You don't even try to understand the problem.
>
> > the other is net, while net maintainer said that is too ealry for rfc.
> >
> > The tested branch is
> > https://github.com/zhangfei-gao/kernel-topics/tree/next-9.23 ( and
> > 9.24-eth), and may keep testing for each cycle.
> > contianing dts, phy, net and all required patchset, assuming one
> > integrated branch will be easier for interal track.
> > The dts will be send later.
> > Testing hardware is ride and rrd.
> >
>
> Explain HOW did you test the bindings in terms of Linux kernel testing.
>
> Then explain WHY you cannot give the tag only to appropriate places,
> instead of blanket one.

Oh, understand now.

I only performed hardware runtime testing,
which did not validate the binding itself.

I should have provided the Tested-by tag only for the driver patches exercised,
rather than applying it to the entire series.

I will run and report dt_binding_check separately for binding patches,
and dtbs_check for the corresponding DTS patches.

Thanks for the guidance, and sorry for the confusion.

Thanks