[PATCH v5] arm64: dts: ti: k3-j784s4-main: Fix mux-reg-masks in serdes_ln_ctrl

[Andrew Davis]

On 2/15/24 10:56 PM, Siddharth Vadapalli wrote:
> On 24/02/13 11:08AM, Andrew Davis wrote:
>> On 2/13/24 3:19 AM, Peter Rosin wrote:
>>> Hi!
>>>
>>> 2024-02-13 at 09:03, Siddharth Vadapalli wrote:
>>>> From: Chintan Vankar <c-vankar at ti.com>
>>>>
>>>> Change offset in mux-reg-masks property for serdes_ln_ctrl node
>>>> since reg-mux property is used in compatible.
>>>>
>>>> Fixes: 2765149273f4 ("mux: mmio: use reg property when parent device is not a syscon")
>>>> Signed-off-by: Chintan Vankar <c-vankar at ti.com>
>>>> Acked-by: Andrew Davis <afd at ti.com>
>>>> Signed-off-by: Siddharth Vadapalli <s-vadapalli at ti.com>
>>>> ---
> ...
> 
>>>> +			mux-reg-masks = <0x0 0x3>, <0x4 0x3>, /* SERDES0 lane0/1 select */
>>>> +					<0x8 0x3>, <0xc 0x3>, /* SERDES0 lane2/3 select */
>>>> +					<0x10 0x3>, <0x14 0x3>, /* SERDES1 lane0/1 select */
>>>> +					<0x18 0x3>, <0x1c 0x3>, /* SERDES1 lane2/3 select */
>>>> +					<0x20 0x3>, <0x24 0x3>, /* SERDES2 lane0/1 select */
>>>> +					<0x28 0x3>, <0x2c 0x3>; /* SERDES2 lane2/3 select */
>>>>    			idle-states = <J784S4_SERDES0_LANE0_PCIE1_LANE0>,
>>>>    				      <J784S4_SERDES0_LANE1_PCIE1_LANE1>,
>>>>    				      <J784S4_SERDES0_LANE2_IP3_UNUSED>,
>>>
>>> Ouch. I suspect there is a similar problem in
>>> arch/arm64/boot/dts/ti/k3-j721e-mcu-wakeup.dtsi:
>>>
>>>
>>> 	fss: bus at 47000000 {
>>> 		compatible = "simple-bus";
>>> 		reg = <0x0 0x47000000 0x0 0x100>;
>>> 		#address-cells = <2>;
>>> 		#size-cells = <2>;
>>> 		ranges;
>>>
>>> 		hbmc_mux: mux-controller at 47000004 {
>>> 			compatible = "reg-mux";
>>> 			reg = <0x00 0x47000004 0x00 0x2>;
>>> 			#mux-control-cells = <1>;
>>> -			mux-reg-masks = <0x4 0x2>; /* HBMC select */
>>> +			mux-reg-masks = <0x0 0x2>; /* HBMC select */
>>> 		};
>>>
>>> Who knows what non-upstreamed devices and devicetrees are affected?
>>> I guess we need to revert 2765149273f4 ("mux: mmio: use reg property
>>> when parent device is not a syscon") unless someone sees a sane way
>>> to fix this.
>>
>> There are only two in-tree nodes with "reg-mux" with a reg property: the
>> one this patch fixes, and the hbmc_mux you point out, both in TI devices.
>> I'd say it is safe to assume we are the only users, and our non-upstreamed
>> DTs depend on that patch, reverting it would cause more issues for
>> out-of-tree users than just fixing the two broken nodes above.
> 
> Peter,
> 
> Is it alright for this patch to be merged, given Andrew's response above?
> The problem with "hbmc_mux" node that you pointed out above could be fixed
> by another patch. Please let me know.

The hbmc_mux fix is now also posted:

https://lore.kernel.org/linux-arm-kernel/20240215141957.13775-1-afd@ti.com/

Andrew

> 
> Regards,
> Siddharth.