Re: [PATCH v2 2/3] mm: drop pte_clear_not_present_full()

On 6/29/26 19:22, Andrew Morton wrote:
> On Mon, 29 Jun 2026 15:49:48 +0200 "David Hildenbrand (Arm)" <[email protected]> wrote:
> 
>> In general, there is no good reason to do anything special when clearing
>> non-present PTEs.
>>
>> In theory, HW that does have to invalidate TLBs for non-present PTEs could
>> benefit from a "full" parameter, but fortunately
>> pte_clear_not_present_full() is not wired up anymore ... and there would
>> have to be something very convincing for us to care about that to re-add
>> it.
>>
>> So, let's just use pte_clear() directly now. To avoid the compiler
>> complaining on some configs about unused "addr" parameter, silence that
>> here.
> 
> Wait, which configs do that?
> 
>> @@ -1022,8 +1007,10 @@ static inline void pte_clear_not_present_full(struct mm_struct *mm,
>>  static inline void clear_not_present_full_ptes(struct mm_struct *mm,
>>  		unsigned long addr, pte_t *ptep, unsigned int nr, int full)
>>  {
>> +	(void)addr;
>> +
> 
> We heavily rely on this warning not happening.
> 
> eg, one of thousands:
> 
> static inline bool page_range_contiguous(const struct page *page,
> 		unsigned long nr_pages)
> {
> 	return true;
> }
> 
> So... what's happening here?

"nr_pages" are not modified in the function, so the compile does not complain.
See below.


A private build bot barked at me after v1 for

	arm-linux-gnueabi-gcc
	openrisc-allnoconfig
	um-allmodconfig

For example:

https://lore.kernel.org/all/[email protected]/

All errors (new ones prefixed by >>):

   In file included from include/linux/kasan.h:38,
                    from include/linux/slab.h:264,
                    from lib/test_bitops.c:12:
   include/linux/pgtable.h: In function 'clear_not_present_full_ptes':
>> include/linux/pgtable.h:974:31: error: parameter 'addr' set but not used
[-Werror=unused-but-set-parameter=]
     974 |                 unsigned long addr, pte_t *ptep, unsigned int nr, int
full)
         |                 ~~~~~~~~~~~~~~^~~~
   cc1: all warnings being treated as errors


The problem is that addr is updated (written) in the function but never read.
This becomes visible once pte_clear() is a macro instead of a function.

	arch/arm/include/asm/pgtable.h:#define pte_clear(mm,addr,ptep)
set_pte_ext(ptep, __pte(0), 0)


"(void) addr" silences bots. An alternative would be to find all such macros and
convert them into (assuming inline function is non-trivial)

#define pte_clear(mm,addr,ptep)  ((void)addr, set_pte_ext(ptep, __pte(0), 0))

Something I wanted to avoid for this simple patch here that just removes one
function indirection.

-- 
Cheers,

David