[PATCH] clk: samsung: exynos-clkout: Remove misleading of_match_table/MODULE_DEVICE_TABLE

[Marek Szyprowski]

On 26.04.2024 11:02, Krzysztof Kozlowski wrote:
> On 26/04/2024 07:24, Marek Szyprowski wrote:
>> On 26.04.2024 02:24, William McVicker wrote:
>>> On 04/25/2024, Marek Szyprowski wrote:
>>>> Since commit 9484f2cb8332 ("clk: samsung: exynos-clkout: convert to
>>>> module driver") this driver is instantiated as MFD-cell (matched by
>>>> platform device name) not as a real platform device created by OF code.
>>>> Remove then of_match_table and related MODULE_DEVICE_TABLE to avoid
>>>> further confusion.
>>>>
>>>> Signed-off-by: Marek Szyprowski <m.szyprowski at samsung.com>
>>>> ---
>>>>    drivers/clk/samsung/clk-exynos-clkout.c | 2 --
>>>>    1 file changed, 2 deletions(-)
>>>>
>>>> diff --git a/drivers/clk/samsung/clk-exynos-clkout.c b/drivers/clk/samsung/clk-exynos-clkout.c
>>>> index 503c6f5b20d5..0c7f4e2aa366 100644
>>>> --- a/drivers/clk/samsung/clk-exynos-clkout.c
>>>> +++ b/drivers/clk/samsung/clk-exynos-clkout.c
>>>> @@ -75,7 +75,6 @@ static const struct of_device_id exynos_clkout_ids[] = {
>>>>    		.data = &exynos_clkout_exynos5,
>>>>    	}, { }
>>>>    };
>>>> -MODULE_DEVICE_TABLE(of, exynos_clkout_ids);
>>> I understand these are duplicates of the exynos-pmu driver, but was wondering
>>> if this will impact the exynos-clkout module from getting auto-loaded? Without
>>> the MODULE_DEVICE_TABLE() defined, the aliases won't be created that trigger
>>> udev to load this module and the mfd driver is obviously not going to load it.
>> This driver loaded and matched only against the platform device name
> Matched yes, but "loaded"? As in module loaded? Are you sure? There is
> no MODULE_ALIAS, no platform_id_table with MODULE_DEVICE_TABLE, so with
> this patch all aliases are gone.
>
> The old aliases indeed could have been used to load the clkout, even if
> not used for matching.

Right, the proper solution is to add:

MODULE_ALIAS("platform:exynos-clkout");

I was convinced that the module_platform_driver() macro adds simple 
module alias based on the driver name, but it looks that's not true. 
Thanks for pointing this out. I will send v2 with the above alias then.

Best regards
-- 
Marek Szyprowski, PhD
Samsung R&D Institute Poland