Re: [PATCH v4 0/6] Add PCIe support for Qualcomm Nord platform

[Qiang Yu]

On Mon, Sep 21, 2026 at 09:37:10AM +0200, Konrad Dybcio wrote:
> On 9/21/26 8:30 AM, Qiang Yu wrote:
> > On Mon, Sep 21, 2026 at 11:07:08AM +0530, Krishna Chaitanya Chundru wrote:
> >> This series adds PCIe support for the Qualcomm Nord (SA8797P) platform.
> >>
> >> Nord has a single Gen5-capable QMP PCIe PHY block split into four
> >> independently-clocked and independently-resettable ports (A/B/C/D).
> >> Firmware selects the active lane topology at boot via a TCSR
> >> configuration register, and the same PHY block can appear as a single
> >> x16 port, x8+x8, x8+x4+x4, or x8+x4+x2+x2 depending on that mode. This
> >> is similar in spirit to the Glymur multi-PHY support already posted,
> >> except Nord's PHY registers are programmed by Linux rather than by
> >> firmware, so this series adds the PLL, TXRXZ, PCS and PCS-lane init
> >> tables along with the power-up/power-down/soft-reset sequencing needed
> >> to bring each port up.
> >>
> >> The series also documents and wires up the qcom,nord-pcie compatible
> >> (reusing the existing SA8775P PCIe controller binding), and adds a
> >> CORE_CLK_CGC_DIS/AUX_PWR_DET workaround required to avoid a timing
> >> convergence issue in the PCIe power domain.
> >>
> >> Signed-off-by: Krishna Chaitanya Chundru <krishna.chundru@xxxxxxxxxxxxxxxx>
> > 
> > Tested on Glymur CRD and Glymur QCB, devices can be enumerated.
> 
> A test of "this didn't break my completely different board" is
> welcome, but looks a little confusing without you saying that
> was the intent specifically - unless maybe it wasn't?
>

Yes, I only confirmed that there is no regression on Glymur.

- Qiang Yu