[PATCH v2 0/7] Add generic PHY driver used by MACB/GEM on EyeQ5

[Philipp Zabel]

On Sa, 2025-11-01 at 09:53 +0100, Théo Lebrun wrote:
> EyeQ5 SoCs integrate two GEM instances. A system-controller register
> region named "OLB" has some control over the Ethernet PHY integration.
> 
> Past iterations [0] touched those syscon registers directly from MACB.
> It was a bad idea. Extend the current OLB ecosystem with a new generic
> PHY driver.
>  - OLB is carried by one main platform driver: clk-eyeq.
>  - It instantiates auxiliary devices: reset-eyeq & pinctrl-eyeq5.
>  - We add a new one: phy-eyeq5-eth.
> 
> Here is a DT overview:
> 
>    olb: system-controller at e00000 {
>            compatible = "mobileye,eyeq5-olb", "syscon";
>            reg = <0 0xe00000 0x0 0x400>;
>            // ...
>            #reset-cells = <2>;
>            #clock-cells = <1>;
>            #phy-cells = <1>; // <= this is new
>    };
> 
>    macb0: ethernet at 2a00000 {
>            compatible = "mobileye,eyeq5-gem";
>            phys = <&olb 0>; // <= GEM device consumes the PHY
>            // ...
>    };
> 
>    macb1: ethernet at 2b00000 {
>            compatible = "mobileye,eyeq5-gem";
>            phys = <&olb 1>; // <= same thing for the second instance
>            // ...
>    };
> 
> The Linux MACB driver already consumes a generic PHY for some other
> compatibles, this is nothing new. The MACB series [1] has been merged
> into net-next/main.
> 
> --
> 
> One topic to talk about: the whole "we must assign child->of_node
> manually". Auxiliary driver core does not do it automatically, so
> either the parent (clk-eyeq) or the children must do it.
> 
> In OLB land, until now, children were doing it with a
> device_set_of_node_from_dev(dev, dev->parent) call in probe.
> 
> Recently, Jerome Brunet added devm_auxiliary_device_create():
> eaa0d30216c1 ("driver core: auxiliary bus: add device creation
> helpers"). Using that cleans up clk-eyeq but means we must remove
> device_set_of_node_from_dev() from reset-eyeq in the same patch series,
> as the helpers do the dev->of_node assignement from the parent driver.
> 
> That explains why the ideal patch:
> [PATCH 4/7] clk: eyeq: add EyeQ5 children auxiliary device for generic PHYs
> Turned into those three:
> [PATCH 3/7] clk: eyeq: use the auxiliary device creation helper
> [PATCH 4/7] clk: eyeq: add EyeQ5 children auxiliary device for generic PHYs
> [PATCH 5/7] reset: eyeq: drop device_set_of_node_from_dev() done by parent
> 
> --
> 
> About merging, it'll probably be complex. I see no build dependencies,
> but the board will be in an odd state if only some patches are applied.
> Some dev_warn() at boot and dev->of_node refcounting issues at unload.
> 
>  - [PATCH 1/7] dt-bindings: soc: mobileye: OLB is an Ethernet PHY provider on EyeQ5
>    We touch dt-bindings because OLB becomes a PHY provider.
>    => linux-mips (?)
> 
>  - [PATCH 2/7] phy: Add driver for EyeQ5 Ethernet PHY wrapper
>    We add the generic PHY driver in drivers/phy/phy-eyeq5-eth.c with the
>    usual Kconfig, Makefile and MAINTAINERS changes.
>    => linux-phy (?)
> 
>  - [PATCH 6/7] MIPS: mobileye: eyeq5: add two Cadence GEM Ethernet controllers
>    [PATCH 7/7] MIPS: mobileye: eyeq5-epm: add two Cadence GEM Ethernet PHYs
>    DTS patches to add both the #phy-cells of OLB and the MACB instances.
>    => linux-mips
> 
>  - [PATCH 4/7] clk: eyeq: add EyeQ5 children auxiliary device for generic PHYs
>    We must update clk-eyeq because it instantiates a new auxdev.
>    => linux-clk
> 
>  - [PATCH 3/7] clk: eyeq: use the auxiliary device creation helper
>    [PATCH 5/7] reset: eyeq: drop device_set_of_node_from_dev() done by parent
>    With the dev->of_node assignement, we must also correct reset-eyeq.
>    => separate them into linux-clk and linux-reset?

Since 3 and 4 should go via clk, and 5 has a dependency on 3, I would
suggest merging them all together.

regards
Philipp