Re: [PATCH v2] ARM: imx: allow to disable board specific PHY fixups

[Florian Fainelli]

On 3/31/2020 11:33 PM, Oleksij Rempel wrote:
> On Tue, Mar 31, 2020 at 09:19:18AM +0100, Russell King - ARM Linux admin wrote:
>> On Tue, Mar 31, 2020 at 10:00:12AM +0200, Marc Kleine-Budde wrote:
>>> On 3/31/20 9:54 AM, Russell King - ARM Linux admin wrote:
>>>> On Tue, Mar 31, 2020 at 09:47:19AM +0200, Marc Kleine-Budde wrote:
>>>>> On 3/30/20 7:41 PM, Russell King - ARM Linux admin wrote:
>>>>>>>> arch/arm/mach-imx/mach-imx6q.c:167:		phy_register_fixup_for_uid(PHY_ID_KSZ9021, MICREL_PHY_ID_MASK,
>>>>>>>> arch/arm/mach-imx/mach-imx6q.c:169:		phy_register_fixup_for_uid(PHY_ID_KSZ9031, MICREL_PHY_ID_MASK,
>>>>>>>> arch/arm/mach-imx/mach-imx6q.c:171:		phy_register_fixup_for_uid(PHY_ID_AR8031, 0xffffffef,
>>>>>>>> arch/arm/mach-imx/mach-imx6q.c:173:		phy_register_fixup_for_uid(PHY_ID_AR8035, 0xffffffef,
>>>>>>
>>>>>> As far as I'm concerned, the AR8035 fixup is there with good reason.
>>>>>> It's not just "random" but is required to make the AR8035 usable with
>>>>>> the iMX6 SoCs.  Not because of a board level thing, but because it's
>>>>>> required for the AR8035 to be usable with an iMX6 SoC.
>>>>>
>>>>> Is this still ture, if the AR8035 is attached to a switch behind an iMX6?
>>>>
>>>> Do you know of such a setup, or are you talking about theoretical
>>>> situations?
>>>
>>> Granted, not for the AR8035, but for one of the KSZ Phys. This is why
>>> Oleksij started looking into this issue in the first place.
>>
>> Maybe there's an easy solution to this - check whether the PHY being
>> fixed up is connected to the iMX6 SoC:
>>
>> static bool phy_connected_to(struct phy_device *phydev,
>> 			     const struct of_device_id *matches)
>> {
>> 	struct device_node *np, *phy_np;
>>
>> 	for_each_matching_node(np, matches) {
>> 		phy_np = of_parse_phandle(np, "phy-handle", 0);
>> 		if (!phy_np)
>> 			phy_np = of_parse_phandle(np, "phy", 0);
>> 		if (!phy_np)
>> 			phy_np = of_parse_phandle(np, "phy-device", 0);
>> 		if (phy_np && phydev->mdio.dev.of_node == phy_np) {
>> 			of_node_put(phy_np);
>> 			of_node_put(np);
>> 			return true;
>> 		}
>> 		of_node_put(phy_np);
>> 	}
>> 	return false;
>> }
>>
>> static struct of_device_id imx_fec_ids[] = {
>> 	{ .compatible = "fsl,imx6q-fec", },
>> 	...
>> 	{ },
>> };
>>
>> static bool phy_connected_to_fec(struct phy_device *phydev)
>> {
>> 	return phy_connected_to(phydev, imx_fec_ids);
>> }
>>
>> and then in the fixups:
>>
>> 	if (!phy_connected_to_fec(phydev))
>> 		return 0;
>>
> 
> Ok, i see. We will limit fixup impact to some specific devicetree nodes.
> And if we wont to disable fixup completely, some special devicetree binding will
> be needed. Correct? Is this acceptable mainline way?
> For the usb ethernet fixups we will need some thing similar.

It looks like IMX is using phy_register_fixup_for_uid() which is not
able to scope the fixup against a specific MDIO bus controller name, I
would suggest we introduce one or two variants of that function in order
to allow specifying the scope against a MDIO bus controller name, and
another variant which can take a comparison function, such that the
logic that Russell has suggested above could be passed a as callback to
a new function: phy_register_fixup_cmp() or whatever is an appropriate
name. Internally, those functions would ideal all be implemented by the
same core function which is able to use any key/value as a match.
-- 
Florian