Re: [PATCH] x86+Arm: drop (rename) __virt_to_maddr() / __maddr_to_virt()

[Jan Beulich <jbeulich@xxxxxxxx>]

On 05.03.2024 20:24, Julien Grall wrote:
> Hi Jan,
> 
> The title is quite confusing. I would have expected the macro...
> 
> On 05/03/2024 08:33, Jan Beulich wrote:
>> There's no use of them anymore except in the definitions of the non-
>> underscore-prefixed aliases. Rename the inline functions, adjust the
>> virt_to_maddr() #define-e, and purge the (x86-only) maddr_to_virt() one,
>> thus eliminating a bogus cast which would have allowed the passing of a
>> pointer type variable into maddr_to_virt() to go silently.
>>
>> Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx>
>>
>> --- a/xen/arch/arm/include/asm/mm.h
>> +++ b/xen/arch/arm/include/asm/mm.h
>> @@ -256,12 +256,12 @@ static inline void __iomem *ioremap_wc(p
>>   /* Page-align address and convert to frame number format */
>>   #define paddr_to_pfn_aligned(paddr)    paddr_to_pfn(PAGE_ALIGN(paddr))
>>   
>> -static inline paddr_t __virt_to_maddr(vaddr_t va)
>> +static inline paddr_t virt_to_maddr(vaddr_t va)
>>   {
>>       uint64_t par = va_to_par(va);
>>       return (par & PADDR_MASK & PAGE_MASK) | (va & ~PAGE_MASK);
>>   }
>> -#define virt_to_maddr(va)   __virt_to_maddr((vaddr_t)(va))
>> +#define virt_to_maddr(va) virt_to_maddr((vaddr_t)(va))
> 
> ... to be removed. But you keep it and just overload the name. I know it 
> is not possible to remove the macro because some callers are using 
> pointers (?).

Indeed. I actually tried without, but the build fails miserably.

> So I would rather prefer if we keep the name distinct on Arm.
> 
> Let see what the other Arm maintainers think.

Well, okay. I'm a little surprised though; I was under the impression
that a goal would be to, eventually, get rid of all the __-prefixed
secondary variants of this kind of helpers.

Jan