diff mbox series

[v4,06/36] mm: Introduce offset_in_thp

Message ID 20200515131656.12890-7-willy@infradead.org (mailing list archive)
State New, archived
Headers show
Series Large pages in the page cache | expand

Commit Message

Matthew Wilcox May 15, 2020, 1:16 p.m. UTC 
From: "Matthew Wilcox (Oracle)" <willy@infradead.org>

Mirroring offset_in_page(), this gives you the offset within this
particular page, no matter what size page it is.  It optimises down
to offset_in_page() if CONFIG_TRANSPARENT_HUGEPAGE is not set.

Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
---
 include/linux/mm.h | 1 +
 1 file changed, 1 insertion(+)

Comments

David Hildenbrand May 15, 2020, 1:39 p.m. UTC | #1 
On 15.05.20 15:16, Matthew Wilcox wrote:
> From: "Matthew Wilcox (Oracle)" <willy@infradead.org>
> 
> Mirroring offset_in_page(), this gives you the offset within this
> particular page, no matter what size page it is.  It optimises down
> to offset_in_page() if CONFIG_TRANSPARENT_HUGEPAGE is not set.
> 
> Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
> ---
>  include/linux/mm.h | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/include/linux/mm.h b/include/linux/mm.h
> index 088acbda722d..9a55dce6a535 100644
> --- a/include/linux/mm.h
> +++ b/include/linux/mm.h
> @@ -1577,6 +1577,7 @@ static inline void clear_page_pfmemalloc(struct page *page)
>  extern void pagefault_out_of_memory(void);
>  
>  #define offset_in_page(p)	((unsigned long)(p) & ~PAGE_MASK)
> +#define offset_in_thp(page, p)	((unsigned long)(p) & (thp_size(page) - 1))
>  
>  /*
>   * Flags passed to show_mem() and show_free_areas() to suppress output in
> 

Reviewed-by: David Hildenbrand <david@redhat.com>
Kirill A . Shutemov May 22, 2020, 5:15 p.m. UTC | #2 
On Fri, May 15, 2020 at 06:16:26AM -0700, Matthew Wilcox wrote:
> From: "Matthew Wilcox (Oracle)" <willy@infradead.org>
> 
> Mirroring offset_in_page(), this gives you the offset within this
> particular page, no matter what size page it is.  It optimises down
> to offset_in_page() if CONFIG_TRANSPARENT_HUGEPAGE is not set.
> 
> Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
> ---
>  include/linux/mm.h | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/include/linux/mm.h b/include/linux/mm.h
> index 088acbda722d..9a55dce6a535 100644
> --- a/include/linux/mm.h
> +++ b/include/linux/mm.h
> @@ -1577,6 +1577,7 @@ static inline void clear_page_pfmemalloc(struct page *page)
>  extern void pagefault_out_of_memory(void);
>  
>  #define offset_in_page(p)	((unsigned long)(p) & ~PAGE_MASK)
> +#define offset_in_thp(page, p)	((unsigned long)(p) & (thp_size(page) - 1))

Looks like thp_mask() would be handy here.
Matthew Wilcox May 29, 2020, 12:59 p.m. UTC | #3 
On Fri, May 22, 2020 at 08:15:17PM +0300, Kirill A. Shutemov wrote:
> On Fri, May 15, 2020 at 06:16:26AM -0700, Matthew Wilcox wrote:
> > +#define offset_in_thp(page, p)	((unsigned long)(p) & (thp_size(page) - 1))
> 
> Looks like thp_mask() would be handy here.

It's not the only place we could use a thp_mask(), but PAGE_MASK is the
inverse of what I think it should be:

include/asm-generic/page.h:#define PAGE_MASK    (~(PAGE_SIZE-1))

ie addr & PAGE_MASK returns the address aligned to page size, not the
offset within the page.  Given this ambiguity, I'm inclined to leave
it as (thp_size(page) - 1), as it's clear which bits we're masking off.
diff mbox series

Patch

diff --git a/include/linux/mm.h b/include/linux/mm.h
index 088acbda722d..9a55dce6a535 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -1577,6 +1577,7 @@  static inline void clear_page_pfmemalloc(struct page *page)
 extern void pagefault_out_of_memory(void);
 
 #define offset_in_page(p)	((unsigned long)(p) & ~PAGE_MASK)
+#define offset_in_thp(page, p)	((unsigned long)(p) & (thp_size(page) - 1))
 
 /*
  * Flags passed to show_mem() and show_free_areas() to suppress output in