Re: [RFC PATCH 0/1] Portable Device Tree Connector -- conceptual

From: David Gibson
Date: Thu Jul 07 2016 - 20:44:11 EST


On Sat, Jul 02, 2016 at 04:55:49PM -0700, frowand.list@xxxxxxxxx wrote:
> From: Frank Rowand <frank.rowand@xxxxxxxxxxx>
>
> Hi All,
>
> This is version 2 of this email.
>
> Changes from version 1:
>
> - some rewording of the text
> - removed new (theoretical) dtc directive "/connector/"
> - added compatibility between mother board and daughter board
> - added info on applying a single .dtbo to different connectors
> - attached an RFC patch showing the required kernel changes
> - changes to mother board .dts connector node:
> - removed target_path property
> - added connector-socket property
> - changes to daughter board .dts connector node:
> - added connector-plug property
>
>
> I've been trying to wrap my head around what Pantelis and Rob have written
> on the subject of a device tree representation of a connector for a
> daughter board to connect to (eg a cape or a shield) and the representation
> of the daughter board. (Or any other physically pluggable object.)
>
> After trying to make sense of what had been written (or presented via slides
> at a conference - thanks Pantelis!), I decided to go back to first principals
> of what we are trying to accomplish. I came up with some really simple bogus
> examples to try to explain what my thought process is.
>
> This is an extremely simple example to illustrate the concepts. It is not
> meant to represent the complexity of a real board.
>
> To start with, assume that the device that will eventually be on a daughter
> board is first soldered onto the mother board. The mother board contains
> two devices connected via bus spi_1. One device is described in the .dts
> file, the other is described in an included .dtsi file.
> Then the device tree files will look like:
>
> $ cat board.dts
> /dts-v1/;
>
> / {
> #address-cells = < 1 >;
> #size-cells = < 1 >;
>
> tree_1: soc@0 {
> reg = <0x0 0x0>;
>
> spi_1: spi1 {
> };
> };
>
> };
>
> &spi_1 {
> ethernet-switch@0 {
> compatible = "micrel,ks8995m";
> };
> };
>
> #include "spi_codec.dtsi"
>
>
> $ cat spi_codec.dtsi
> &spi_1 {
> codec@1 {
> compatible = "ti,tlv320aic26";
> };
> };
>
>
> #----- codec chip on cape
>
> Then suppose I move the codec chip to a cape. Then I will have the same
> exact .dts and .dtsi and everything still works.
>
>
> @----- codec chip on cape, overlay
>
> If I want to use overlays, I only have to add the version and "/plugin/",
> then use the '-@' flag for dtc (both for the previous board.dts and
> this spi_codec_overlay.dts):
>
> $ cat spi_codec_overlay.dts
> /dts-v1/;
>
> /plugin/;
>
> &spi_1 {
> codec@1 {
> compatible = "ti,tlv320aic26";
> };
> };
>
>
> Pantelis pointed out that the syntax has changed to be:
> /dts-v1/ /plugin/;
>
>
> #----- codec chip on cape, overlay, connector
>
> Now we move into the realm of connectors. My mental model of what the
> hardware and driver look like has not changed. The only thing that has
> changed is that I want to be able to specify that the connector that
> the cape is plugged into has some pins that are the spi bus /soc/spi1.
>
> The following _almost_ but not quite gets me what I want. Note that
> the only thing the connector node does is provide some kind of
> pointer or reference to what node(s) are physically routed through
> the connector. The connector node does not need to describe the pins;
> it only has to point to the node that describes the pins.
>
> This example will turn out to be not sufficient. It is a stepping
> stone in building my mental model.
>
> $ cat board_with_connector.dts
> /dts-v1/;
>
> / {
> #address-cells = < 1 >;
> #size-cells = < 1 >;
>
> tree_1: soc@0 {
> reg = <0x0 0x0>;
>
> spi_1: spi1 {
> };
> };
>
> connector_1: connector_1 {
> spi1 {
> target_phandle = <&spi_1>;
> };
> };
>
> };
>
> &spi_1 {
> ethernet-switch@0 {
> compatible = "micrel,ks8995m";
> };
> };
>
>
> $ cat spi_codec_overlay_with_connector.dts
> /dts-v1/;
>
> /plugin/;
>
> &connector_1 {
> spi1 {
> codec@1 {
> compatible = "ti,tlv320aic26";
> };
> };
> };
>
>
> The result is that the overlay fixup for spi1 on the cape will
> relocate the spi1 node to /connector_1 in the host tree, so
> this does not solve the connector linkage yet:
>
> -- chunk from the decompiled board_with_connector.dtb:
>
> __symbols__ {
> connector_1 = "/connector_1";
> };
>
> -- chunk from the decompiled spi_codec_overlay_with_connector.dtb:
>
> fragment@0 {
> target = <0xffffffff>;
> __overlay__ {
> spi1 {
> codec@1 {
> compatible = "ti,tlv320aic26";
> };
> };
> };
> };
> __fixups__ {
> connector_1 = "/fragment@0:target:0";
> };
>
>
> After applying the overlay, the codec@1 node will be at
> /connector_1/spi1/codec@1. What I want is for that node
> to be at /spi1/codec@1.
>
>
>
> #----- magic new syntax
>
> What I really want is some way to tell dtc that I want to do one
> level of dereferencing when resolving the path of device nodes
> contained by the connector node in the overlay dts.
>
> Version 1 of this email suggested using dtc magic to do this extra
> level of dereferencing. This version of the email has changed to
> have the kernel code that applies the overlay do the extra level
> of dereferencing.
>
> The property "connector-socket" tells the kernel overlay code
> that this is a socket. The overlay code does not actually
> do anything special as a result of this property; it is simply
> used as a sanity check that this node really is a socket. The
> person writing the mother board .dts must provide the
> target_phandle property, which points to a node responsible for
> some of the pins on the connector.
>
> The property "connector-plug" tells the kernel overlay code
> that each child node in the overlay corresponds to a node in the
> socket, and the socket will contain one property that is
> a phandle pointing to the node that is the target of that child
> node in the overlay node.
>
>
> $ cat board_with_connector_v2.dts
>
> /dts-v1/;
>
> / {
> #address-cells = < 1 >;
> #size-cells = < 1 >;
>
> tree_1: soc@0 {
> reg = <0x0 0x0>;
>
> spi_1: spi1 {
> };
> };
>
> connector_1: connector_1 {
> compatible = "11-pin-accessory";
> connector-socket;

I don't see any advantage to allowing connectors anywhere in the tree:
pretty much by definition a connector is a "whole board" concept. So
I think instead they should all go in a new special node under the
root, say /connectors. With that done, you don't need the
connector-socket tag any more.

> spi1 {
> target_phandle = <&spi_1>;
> };
> };
>
> };
>
> &spi_1 {
> ethernet-switch@0 {
> compatible = "micrel,ks8995m";
> };
> };
>
>
> $ cat spi_codec_overlay_with_connector_v2.dts
>
> /dts-v1/;
>
> /plugin/;
>
> &connector_1 {
> connector-plug;
> compatible = "11-pin-accessory";
>
> spi1 {
> codec@1 {
> compatible = "ti,tlv320aic26";
> };
> };
> };
>
>
> The spi_codec_overlay_with_connector_v2.dtb __fixups__ information
> is unchanged from the previous example, but the kernel overlay
> code will do the correct extra level of dereferencing when it
> detects the connector-plug property in the overlay.
>
> The one remaining piece that this patch does not provide is how
> the overlay manager (which does not yet exist in the mainline
> tree) can apply an overlay to two different targets. That
> final step should be a trivial change to of_overlay_create(),
> adding a parameter that is a mapping of the target (or maybe
> even targets) in the overlay to different targets in the
> active device tree.
>
> This seems like a more straight forward way to handle connectors.
>
> First, ignoring pinctrl and pinmux, what does everyone think?
>
> Then, the next step is whether pinctrl and pinmux work with this method.
> Pantelis, can you point me to a good example for
>
> 1) an in-tree board dts file
> 2) an overlay file (I am assuming out of tree) that applies to the board
> 3) an in-tree .dtsi file that would provide the same features as
> the overlay file if it was included by the board dts file
>
> It should be easier to discuss pinctrl and pinmux with an example.

Hrm.. so I think you're trying to stick too close to the existing
overlay model. Something I've always disliked about that model is
that the plugin can overlay *anywhere* in the master tree, meaning it
must have intimate knowledge of that tree. Instead of using the
global __symbols__, there should be a set of "symbols" local to the
specific connector (socket), which are the *only* points which the
plugin is allowed to overlay or reference.

Given that we're going to need new code to support this new connector
model, I think we should also fix some of the uglies in the current
overlay format while we're at it.

I have to run now, but I'll try to send out a counter-proposal
shortly.

--
David Gibson | I'll have my music baroque, and my code
david AT gibson.dropbear.id.au | minimalist, thank you. NOT _the_ _other_
| _way_ _around_!
http://www.ozlabs.org/~dgibson

Attachment: signature.asc
Description: PGP signature