Re: [PATCH] mm/pagewalk: Clean up folio_walk_start() by removing expose_page

[David Hildenbrand]

On 12.06.25 14:09, peng li wrote:
> The name expose_page is confusing to understand. From the code logic, it is
> probably meant to express "belongs to the normal page and get offset from
> the mapped page"?
> Perhaps changing "expose_page" to "normal_mapped_page" can better express
> its behavior? But perhaps its existence is meaningless, because fw->page
> can be directly confirmed and obtained from the page type without the need
> for a separate flag.
>
> Key changes:
> 1. Remove expose_page and its conditional logic
> 2. Always set fw->page when a valid page is found
> 3. Add clarifying comments about offset calculation
> 4. Initialize fw->page to NULL at PMD/PTE levels
>
> Signed-off-by: peng li <peng8420.li@xxxxxxxxx>
> ---


I don't see the cleanup here, really.


--
Cheers,

David / dhildenb