2024-05-27 04:45:48

by Oscar Salvador

Subject: [PATCH] arch/x86: Do not explicitly clear Reserved flag in free_pagetable

In free_pagetable() we use the non-atomic version for clearing the
PageReserved bit from the page.
free_pagetable() will either call free_reserved_page() or
put_page_bootmem(), which will eventually end up calling
free_serverd_page(), and in there we already clear the PageReserved flag.

Signed-off-by: Oscar Salvador <[email protected]>
---
This has been like this since commit ae9aae9eda2d ("memory-hotplug: common APIs to
support page tables hot-remove"), so I might be missing something obvious, but
I cannot explain why we would need that __ClearPageReserved upfront, when
free_reserved_pages() already does that for us.
---
arch/x86/mm/init_64.c | 2 --
1 file changed, 2 deletions(-)

diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c
index 7e177856ee4f..9f4778c2cf9f 100644
--- a/arch/x86/mm/init_64.c
+++ b/arch/x86/mm/init_64.c
@@ -980,8 +980,6 @@ static void __meminit free_pagetable(struct page *page, int order)

/* bootmem page has reserved flag */
if (PageReserved(page)) {
- __ClearPageReserved(page);
-
magic = page->index;
if (magic == SECTION_INFO || magic == MIX_SECTION_INFO) {
while (nr_pages--)
--
2.45.1

2024-05-27 07:46:50

by David Hildenbrand

Subject: Re: [PATCH] arch/x86: Do not explicitly clear Reserved flag in free_pagetable

Am 27.05.24 um 06:45 schrieb Oscar Salvador:
> In free_pagetable() we use the non-atomic version for clearing the
> PageReserved bit from the page.
> free_pagetable() will either call free_reserved_page() or
> put_page_bootmem(), which will eventually end up calling
> free_serverd_page(), and in there we already clear the PageReserved flag.
>
> Signed-off-by: Oscar Salvador <[email protected]>
> ---
> This has been like this since commit ae9aae9eda2d ("memory-hotplug: common APIs to
> support page tables hot-remove"), so I might be missing something obvious, but
> I cannot explain why we would need that __ClearPageReserved upfront, when
> free_reserved_pages() already does that for us.

Especially, we only call it on the first page (in case we'd have order>0).

Acked-by: David Hildenbrand <[email protected]>

--
Thanks,

David / dhildenb