Re: [PATCH v2 06/26] xen/riscv: implement make_cpus_node()

[Jan Beulich <jbeulich@suse.com>]

On 19.05.2026 15:33, Oleksii Kurochko wrote:
> On 5/18/26 6:00 PM, Jan Beulich wrote:
>> On 08.05.2026 16:43, Oleksii Kurochko wrote:
>>> +    {
>>> +        char buf[64];
>>> +        uint32_t reg = cpu_to_fdt32(cpu);
>>
>> Isn't this a byte-order adjustment? If so, how come ...
>>
>>> +        snprintf(buf, sizeof(buf), "cpu@%u", cpu);
>>
>> ... the result is passed to an entirely non-FDT function? (Most pre-existing uses
>> of the function that I can spot store something in memory, i.e. adjusting byte-
>> order makes sense there.)
> 
> But here pure cpu is used instead of reg variable and reg variable is 
> used here ...
> 
>>
>>> +        res = fdt_begin_node(fdt, buf);
>>> +        if ( res )
>>> +            return res;
>>> +
>>> +        res = fdt_property(fdt, "reg", &reg, sizeof(reg));
>>> +        if ( res )
>>> +            return res;
> ... but it we could drop it and use just:
>    res = fdt_property_cell(fdt, "reg", cpu);
> 
> fdt_property_cell will take care if a byte-order adjustment.

I'm sorry, I clearly must have seen something that wasn't really there.

Jan