Re: [PATCH v3 4/4] soc: sunxi: sram: Only iterate over SRAM children

[Samuel Holland <samuel@sholland.org>]

On 8/16/22 5:03 AM, Krzysztof Kozlowski wrote:
> On 16/08/2022 13:01, Krzysztof Kozlowski wrote:
>> On 15/08/2022 07:34, Samuel Holland wrote:
>>> Now that a "regulators" child is accepted by the controller binding, the
>>> debugfs show routine must be explicitly limited to "sram" children.
>>>
>>> Signed-off-by: Samuel Holland <samuel@sholland.org>
>>> ---
>>>
>>> (no changes since v2)
>>>
>>> Changes in v2:
>>>  - New patch for v2
>>>
>>>  drivers/soc/sunxi/sunxi_sram.c | 3 +++
>>>  1 file changed, 3 insertions(+)
>>>
>>> diff --git a/drivers/soc/sunxi/sunxi_sram.c b/drivers/soc/sunxi/sunxi_sram.c
>>> index 92f9186c1c42..6acaaeb65652 100644
>>> --- a/drivers/soc/sunxi/sunxi_sram.c
>>> +++ b/drivers/soc/sunxi/sunxi_sram.c
>>> @@ -120,6 +120,9 @@ static int sunxi_sram_show(struct seq_file *s, void *data)
>>>  	seq_puts(s, "--------------------\n\n");
>>>  
>>>  	for_each_child_of_node(sram_dev->of_node, sram_node) {
>>> +		if (!of_node_name_eq(sram_node, "sram"))
>>
>> You should not rely on node names. They can change in DTS. Why do you
>> need to test for the name?
>>
> 
> Ah, it is not a device node but a child property, right? In such case,
> it's of course fine.

It is a child node.

> The device node names could change and should not be considered ABI (at
> least I hope should not...).

The node names are limited by patternProperties in the controller binding. I can
check the child nodes for compatibility with "mmio-sram" if that is better.

Regards,
Samuel