Re: [PATCH 2/2] mm/mm.h: Write folio->_flags_1 & 0xff as a macro definition

[Dev Jain <dev.jain@arm.com>]

On 12/02/25 12:37 pm, liuye wrote:
> 
> 
> 在 2025/2/12 13:12, Dev Jain 写道:
>>
>>
>> On 12/02/25 8:28 am, Liu Ye wrote:
>>> There are multiple locations in mm.h where (folio->_flags_1 & 0xff) is
>>> used. Write it as a macro definition to improve the readability and
>>> maintainability of the code.
>>>
>>> Signed-off-by: Liu Ye <liuye@kylinos.cn>
>>> ---
>>>    include/linux/mm.h | 10 ++++++----
>>>    1 file changed, 6 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/include/linux/mm.h b/include/linux/mm.h
>>> index 7b1068ddcbb7..750e75f45557 100644
>>> --- a/include/linux/mm.h
>>> +++ b/include/linux/mm.h
>>> @@ -1098,6 +1098,8 @@ int vma_is_stack_for_current(struct vm_area_struct *vma);
>>>    struct mmu_gather;
>>>    struct inode;
>>>    +#define FOLIO_ORDER(folio) ((folio)->_flags_1 & 0xff)
>>> +
>>>    /*
>>>     * compound_order() can be called without holding a reference, which means
>>>     * that niceties like page_folio() don't work.  These callers should be
>>> @@ -1111,7 +1113,7 @@ static inline unsigned int compound_order(struct page *page)
>>>          if (!test_bit(PG_head, &folio->flags))
>>>            return 0;
>>> -    return folio->_flags_1 & 0xff;
>>> +    return FOLIO_ORDER(folio);
>>>    }
>>>      /**
>>> @@ -1127,7 +1129,7 @@ static inline unsigned int folio_order(const struct folio *folio)
>>>    {
>>>        if (!folio_test_large(folio))
>>>            return 0;
>>> -    return folio->_flags_1 & 0xff;
>>> +    return FOLIO_ORDER(folio);
>>>    }
>>>      #include <linux/huge_mm.h>
>>> @@ -2061,7 +2063,7 @@ static inline long folio_nr_pages(const struct folio *folio)
>>>    #ifdef CONFIG_64BIT
>>>        return folio->_folio_nr_pages;
>>>    #else
>>> -    return 1L << (folio->_flags_1 & 0xff);
>>> +    return 1L << FOLIO_ORDER(folio);
>>>    #endif
>>>    }
>>>    @@ -2086,7 +2088,7 @@ static inline unsigned long compound_nr(struct page *page)
>>>    #ifdef CONFIG_64BIT
>>>        return folio->_folio_nr_pages;
>>>    #else
>>> -    return 1L << (folio->_flags_1 & 0xff);
>>> +    return 1L << FOLIO_ORDER(folio);
>>>    #endif
>>>    }
>>>    
>>
>> Personally I do not think this is improving readability. You are introducing one more macro for people to decipher instead of directly seeing folio->_flags_1 & 0xff. This is similar to whether to write
>> if (x) => do_stuff(), or if (x != 0) => do_stuff(). The former is more "readable" by convention but the latter makes it easier and obvious to understand.
>>
> Or simply for maintenance purposes, if the meaning of a bit changes, only the macro definition needs to be modified.

Well, then let us wait for that time to come :) Personally I am not a 
fan of over-abstracting, especially when it is just a single line; one 
benefit I have seen of writing the way it is written right now, is that 
I actually get reminded where the folio order is actually stored. I have 
no objection on getting this patch applied, if someone else thinks this 
is a fruitful abstraction. In any case, you do need to come up with a 
better name than FOLIO_ORDER, as it is confusing.

> 
> Thanks，
> Liu Ye
> 

> 
>