Re: [PATCH] powerpc/sysdev: Use of_device_get_match_data()

[Michael Ellerman]

Christophe Leroy <christophe.leroy@xxxxxxxxxx> writes:
> Le 04/03/2022 à 15:26, Marc Zyngier a écrit :
>> On Fri, 04 Mar 2022 13:10:19 +0000,
>> Christophe Leroy <christophe.leroy@xxxxxxxxxx> wrote:
>>> Le 04/03/2022 à 02:18, cgel.zte@xxxxxxxxx a écrit :
>>>> From: Minghao Chi (CGEL ZTE) <chi.minghao@xxxxxxxxxx>
>>>>
>>>> Use of_device_get_match_data() to simplify the code.
>>>>
>>>> Reported-by: Zeal Robot <zealci@xxxxxxxxxx>
>>>> Signed-off-by: Minghao Chi (CGEL ZTE) <chi.minghao@xxxxxxxxxx>
>>>> ---
>>>>    arch/powerpc/sysdev/fsl_msi.c | 6 +-----
>>>>    1 file changed, 1 insertion(+), 5 deletions(-)
>>>>
>>>> diff --git a/arch/powerpc/sysdev/fsl_msi.c b/arch/powerpc/sysdev/fsl_msi.c
>>>> index b3475ae9f236..9d135bbb30b7 100644
>>>> --- a/arch/powerpc/sysdev/fsl_msi.c
>>>> +++ b/arch/powerpc/sysdev/fsl_msi.c
>>>> @@ -387,7 +387,6 @@ static int fsl_msi_setup_hwirq(struct fsl_msi *msi, struct platform_device *dev,
>>>>    static const struct of_device_id fsl_of_msi_ids[];
>>>>    static int fsl_of_msi_probe(struct platform_device *dev)
>>>>    {
>>>> -	const struct of_device_id *match;
>>>>    	struct fsl_msi *msi;
>>>>    	struct resource res, msiir;
>>>>    	int err, i, j, irq_index, count;
>>>> @@ -397,10 +396,7 @@ static int fsl_of_msi_probe(struct platform_device *dev)
>>>>    	u32 offset;
>>>>    	struct pci_controller *phb;
>>>>    
>>>> -	match = of_match_device(fsl_of_msi_ids, &dev->dev);
>>>> -	if (!match)
>>>> -		return -EINVAL;
>>>> -	features = match->data;
>>>> +	features = of_device_get_match_data(&dev->dev);
>>>
>>> What happens when features is NULL ?
>> 
>> I did jump at that one too, but as it turns out, it cannot happen, by
>> construction. All the fsl_of_msi_ids[] entries have a non-NULL .data
>> pointer, and you only enter probe if you match a fsl_of_msi_ids[]
>> entry with the DT.
>> 
>> So the current check for a NULL match is not something that can happen
>> short of some other bug somewhere.
>> 
>
> Ok.
>
> Then it would be good to have a sentence explaining that in the commit 
> message.

Yes I agree. Too many of these cleanup patches assume the reviewer is
intimately familiar with the details of the new API/helper and don't
explain things fully in the change log.

The helper is:

const void *of_device_get_match_data(const struct device *dev)
{
	const struct of_device_id *match;

	match = of_match_device(dev->driver->of_match_table, dev);
	if (!match)
		return NULL;

	return match->data;
}

So as Marc says, if we're in probe then there must be a match.

IMHO there should be a check for match->data being NULL. Although the
current match table contains a non-NULL data for every element, that
could easily change in future (although this driver will probably never
be updated).

The forward declaration of fsl_of_msi_ids should also be removed, as it
will no longer be necessary.

cheers