Re: [PATCH v2 1/4] dt-bindings: connector: Add PCIe M.2 Mechanical Key M connector

[Rob Herring <robh@kernel.org>]

On Wed, Nov 12, 2025 at 10:12:36PM +0200, Dmitry Baryshkov wrote:
> On Tue, Nov 11, 2025 at 07:19:45PM +0530, Manivannan Sadhasivam wrote:
> > On Sun, Nov 09, 2025 at 10:13:59PM +0200, Dmitry Baryshkov wrote:
> > > On Sun, Nov 09, 2025 at 09:48:02PM +0530, Manivannan Sadhasivam wrote:
> > > > On Sat, Nov 08, 2025 at 08:10:54PM +0200, Dmitry Baryshkov wrote:
> > > > > On Sat, Nov 08, 2025 at 08:53:19AM +0530, Manivannan Sadhasivam wrote:
> > > > > > Add the devicetree binding for PCIe M.2 Mechanical Key M connector defined
> > > > > > in the PCI Express M.2 Specification, r4.0, sec 5.3. This connector
> > > > > > provides interfaces like PCIe and SATA to attach the Solid State Drives
> > > > > > (SSDs) to the host machine along with additional interfaces like USB, and
> > > > > > SMB for debugging and supplementary features. At any point of time, the
> > > > > > connector can only support either PCIe or SATA as the primary host
> > > > > > interface.
> > > > > > 
> > > > > > The connector provides a primary power supply of 3.3v, along with an
> > > > > > optional 1.8v VIO supply for the Adapter I/O buffer circuitry operating at
> > > > > > 1.8v sideband signaling.
> > > > > > 
> > > > > > The connector also supplies optional signals in the form of GPIOs for fine
> > > > > > grained power management.
> > > > > > 
> > > > > > Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@oss.qualcomm.com>
> > > > > > ---
> > > > > >  .../bindings/connector/pcie-m2-m-connector.yaml    | 122 +++++++++++++++++++++
> > > > > >  1 file changed, 122 insertions(+)
> > > > > > 
> > > > > > diff --git a/Documentation/devicetree/bindings/connector/pcie-m2-m-connector.yaml b/Documentation/devicetree/bindings/connector/pcie-m2-m-connector.yaml
> > > > > > new file mode 100644
> > > > > > index 0000000000000000000000000000000000000000..be0a3b43e8fd2a2a3b76cad4808ddde79dceaa21
> > > > > > --- /dev/null
> > > > > > +++ b/Documentation/devicetree/bindings/connector/pcie-m2-m-connector.yaml
> > > > > > @@ -0,0 +1,122 @@
> > > > > > +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
> > > > > > +%YAML 1.2
> > > > > > +---
> > > > > > +$id: http://devicetree.org/schemas/connector/pcie-m2-m-connector.yaml#
> > > > > > +$schema: http://devicetree.org/meta-schemas/core.yaml#
> > > > > > +
> > > > > > +title: PCIe M.2 Mechanical Key M Connector
> > > > > > +
> > > > > > +maintainers:
> > > > > > +  - Manivannan Sadhasivam <manivannan.sadhasivam@oss.qualcomm.com>
> > > > > > +
> > > > > > +description:
> > > > > > +  A PCIe M.2 M connector node represents a physical PCIe M.2 Mechanical Key M
> > > > > > +  connector. The Mechanical Key M connectors are used to connect SSDs to the
> > > > > > +  host system over PCIe/SATA interfaces. These connectors also offer optional
> > > > > > +  interfaces like USB, SMB.
> > > > > > +
> > > > > > +properties:
> > > > > > +  compatible:
> > > > > > +    const: pcie-m2-m-connector
> > > > > 
> > > > > Is a generic compatible enough here? Compare this to the USB connectors,
> > > > > which, in case of an independent USB-B connector controlled/ing GPIOs,
> > > > > gets additional gpio-usb-b-connector?
> > > > > 
> > > > 
> > > > I can't comment on it as I've not seen such usecases as of now. But I do think
> > > > that this generic compatible should satisfy most of the design requirements. If
> > > > necessity arises, a custom compatible could be introduced with this generic one
> > > > as a fallback.
> > > 
> > > Ack
> > > 
> > > > 
> > > > > > +
> > > > > > +  vpcie3v3-supply:
> > > > > > +    description: A phandle to the regulator for 3.3v supply.
> > > > > > +
> > > > > > +  vio1v8-supply:
> > > > > > +    description: A phandle to the regulator for VIO 1.8v supply.
> > > > > > +
> > > > > > +  ports:
> > > > > > +    $ref: /schemas/graph.yaml#/properties/ports
> > > > > > +    description: OF graph bindings modeling the interfaces exposed on the
> > > > > > +      connector. Since a single connector can have multiple interfaces, every
> > > > > > +      interface has an assigned OF graph port number as described below.
> > > > > > +
> > > > > > +    properties:
> > > > > > +      port@0:
> > > > > > +        $ref: /schemas/graph.yaml#/properties/port
> > > > > > +        description: PCIe/SATA interface
> > > > > 
> > > > > Should it be defined as having two endpoints: one for PCIe, one for
> > > > > SATA?
> > > > > 
> > > > 
> > > > I'm not sure. From the dtschema of the connector node:
> > > > 
> > > > "If a single port is connected to more than one remote device, an 'endpoint'
> > > > child node must be provided for each link"
> > > > 
> > > > Here, a single port is atmost connected to only one endpoint and that endpoint
> > > > could PCIe/SATA. So IMO, defining two endpoint nodes doesn't fit here.
> > > 
> > > I think this needs to be better defined. E.g. there should be either one
> > > endpoint going to the shared SATA / PCIe MUX, which should then be
> > > controlled somehow, in a platform-specific way (how?) or there should be
> > > two endpoints defined, e.g. @0 for SATA and @1 for PCIe (should we
> > > prevent powering up M.2 if PEDET points out the unsupported function?).
> > > (Note: these questions might be the definitive point for the bare
> > > m2-m-connector vs gpio-m2-m-connector: the former one defines just the
> > > M.2 signals, letting e.g. UEFI or PCIe controller to react to them, the
> > > latter one defines how to control MUXes, the behaviour wrt PEDET, etc.,
> > > performing all those actions in OS driver).
> > > 
> > 
> > In the case of an external GPIO controlled MUX for PCIe/SATA interface, I would
> > assume that the MUX will be controlled by the PEDET itself. PEDET will be driven
> > low by the card if it uses SATA, pulled high (NC) if it uses PCIe. Then that
> > signal will help the MUX to route the proper interface to the connector.
> > 
> > Even in that case, there should be a single endpoint coming from the MUX to the
> > connector.
> 
> How would you model this in the actual DT? We don't have separate
> PCIe/SATA muxes in DT, do we?

Do we have to describe it? On an x86 system, does something have to 
describe what's connected? Wouldn't you try a PCIe link and then a SATA 
link if the PCIe link doesn't come up?

Rob