Re: [RFC 1/2] lib/vsprintf: Add support for pte_t

[Anshuman Khandual <anshuman.khandual@arm.com>]

On 18/06/25 2:13 PM, David Hildenbrand wrote:
> On 18.06.25 10:33, Anshuman Khandual wrote:
>>
>>
>> On 18/06/25 1:49 PM, David Hildenbrand wrote:
>>>> +    case 'p':
>>>> +        if (fmt[1] == 't' && fmt[2] == 'e') {
>>>> +            pte_t *pte = (pte_t *)ptr;
>>>> +
>>>> +            spec.field_width = 10;
>>>> +            spec.precision = 8;
>>>> +            spec.base = 16;
>>>> +            spec.flags = SPECIAL | SMALL | ZEROPAD;
>>>> +            if (sizeof(pte_t) == sizeof(u64)) {
>>>> +                u64 val = pte_val(*pte);
>>>> +
>>>> +                return number(buf, end, val, spec);
>>>> +            }
>>>> +            WARN_ONCE(1, "Non standard pte_t\n");
>>>
>>> What about 32bit with 32bit pte_t?
>>
>> Ahh, missed that. Just wondering which all platforms might
>> care about the 32 bit pte representation.
> 
> I think e.g., 32bit arm has 32bit ptes?
> 
> arch/arm/include/asm/pgtable-2level-types.h
> 
> typedef u32 pteval_t;
> ...
> typedef struct { pteval_t pte; } pte_t;

Right, missed that. I will accommodate 32 bit representations.