Re: [PATCH v3 3/3] EDAC: nuvoton: Add NPCM memory controller driver

[Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>]

On 13/03/2022 21:22, Avi Fishman wrote:
> On Fri, Mar 11, 2022 at 11:15 AM Krzysztof Kozlowski
> <krzysztof.kozlowski@canonical.com> wrote:
>>
>> On 11/03/2022 02:42, Medad CChien wrote:
>>> Add support for Nuvoton NPCM SoC.
>>>
>>> Signed-off-by: Medad CChien <ctcchien@nuvoton.com>
>>> ---
>>>  drivers/edac/Kconfig     |   9 +
>>>  drivers/edac/Makefile    |   1 +
>>>  drivers/edac/npcm_edac.c | 714 +++++++++++++++++++++++++++++++++++++++
>>>  3 files changed, 724 insertions(+)
>>>  create mode 100644 drivers/edac/npcm_edac.c
>>>
>>> diff --git a/drivers/edac/Kconfig b/drivers/edac/Kconfig
>>> index 58ab63642e72..757e1d160640 100644
>>> --- a/drivers/edac/Kconfig
>>> +++ b/drivers/edac/Kconfig
>>> @@ -539,4 +539,13 @@ config EDAC_DMC520
>>>         Support for error detection and correction on the
>>>         SoCs with ARM DMC-520 DRAM controller.
>>>
>>> +config EDAC_NPCM
>>> +     tristate "Nuvoton NPCM DDR Memory Controller"
>>> +     depends on ARCH_NPCM
>>
>> || COMPILE_TEST
>> (and test if it compiles)
>>
>> (...)
>>
>>> +
>>> +MODULE_DEVICE_TABLE(of, npcm_edac_of_match);
>>> +
>>> +static int npcm_edac_mc_probe(struct platform_device *pdev)
>>> +{
>>> +     const struct npcm_edac_platform_data *npcm_chip;
>>> +     struct device *dev = &pdev->dev;
>>> +     struct edac_mc_layer layers[1];
>>> +     const struct of_device_id *id;
>>> +     struct priv_data *priv_data;
>>> +     struct mem_ctl_info *mci;
>>> +     struct resource *res;
>>> +     void __iomem *reg;
>>> +     int ret = -ENODEV;
>>> +     int irq;
>>> +
>>> +     id = of_match_device(npcm_edac_of_match, &pdev->dev);
>>> +     if (!id)
>>> +             return -ENODEV;
>>
>> Why do you need it? How such case is even possible?
>>
>>> +
>>> +     npcm_chip = of_device_get_match_data(&pdev->dev);
>>> +     if (!npcm_chip)
>>> +             return -ENODEV;
>>
>> I wonder, how is it possible to have here NULL?
>>
> Both of_match_device() and of_device_get_match_data() can return NULL,
> are we missing something?

I think your driver is OF-only, right? If yes, how is it possible to be
here in probe() (meaning: being matched by of_device_id) and a few lines
later do not match the same of_device_id?

Best regards,
Krzysztof