Sparsemem: Do not reserve section flags if VMEMMAP is in use

Sparsemem: Do not reserve section flags if VMEMMAP is in use

[Christoph Lameter]

Sparsemem with virtual memmap does not need the section flags in
the page flags. Do not allocate the bits if they are not needed.

This has the potential of freeing up a lot of page flags if SPARSE
can be made to consistently use a virtual memmap.

Signed-off-by: Christoph Lameter <clameter@sgi.com>

---
 include/linux/mm.h |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Index: linux-2.6/include/linux/mm.h
===================================================================
--- linux-2.6.orig/include/linux/mm.h	2007-11-12 19:36:39.472347109 -0800
+++ linux-2.6/include/linux/mm.h	2007-11-12 19:37:05.197064250 -0800
@@ -378,7 +378,7 @@ static inline void set_compound_order(st
  * with space for node: | SECTION | NODE | ZONE | ... | FLAGS |
  *   no space for node: | SECTION |     ZONE    | ... | FLAGS |
  */
-#ifdef CONFIG_SPARSEMEM
+#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
 #define SECTIONS_WIDTH		SECTIONS_SHIFT
 #else
 #define SECTIONS_WIDTH		0

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: Sparsemem: Do not reserve section flags if VMEMMAP is in use

[KAMEZAWA Hiroyuki]

On Mon, 12 Nov 2007 19:47:06 -0800 (PST)
Christoph Lameter <clameter@sgi.com> wrote:

> Index: linux-2.6/include/linux/mm.h
> ===================================================================
> --- linux-2.6.orig/include/linux/mm.h	2007-11-12 19:36:39.472347109 -0800
> +++ linux-2.6/include/linux/mm.h	2007-11-12 19:37:05.197064250 -0800
> @@ -378,7 +378,7 @@ static inline void set_compound_order(st
>   * with space for node: | SECTION | NODE | ZONE | ... | FLAGS |
>   *   no space for node: | SECTION |     ZONE    | ... | FLAGS |
>   */
> -#ifdef CONFIG_SPARSEMEM
> +#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
>  #define SECTIONS_WIDTH		SECTIONS_SHIFT
>  #else
>  #define SECTIONS_WIDTH		0
> 
I like this. but it may safe to add this definition to do this..

==
#if SECTIONS_WIDTH > 0
static inline page_to_section(struct page *page)
{
	return pfn_to_section(page_to_pfn(page));
}
else
....
#endif
==

page_to_section is used in page_to_nid() if NODE_NOT_IN_PAGE_FLAGS=y.
(I'm not sure exact config dependency.)

Thanks,
-Kame

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: Sparsemem: Do not reserve section flags if VMEMMAP is in use

[Christoph Lameter]

On Tue, 13 Nov 2007, KAMEZAWA Hiroyuki wrote:

> I like this. but it may safe to add this definition to do this..
> 
> ==
> #if SECTIONS_WIDTH > 0
> static inline page_to_section(struct page *page)
> {
> 	return pfn_to_section(page_to_pfn(page));
> }
> else
> ....
> #endif
> ==

Well that is currently not done for !SPARSEMEM configuration where 
SECTIONS_WIDTH is also zero. So I left it as is.

> page_to_section is used in page_to_nid() if NODE_NOT_IN_PAGE_FLAGS=y.
> (I'm not sure exact config dependency.)

NODE_NOT_IN_PAGE_FLAGS=y only occurs when flag bits are 
taken away by sparsemem for the section bits.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: Sparsemem: Do not reserve section flags if VMEMMAP is in use

[KAMEZAWA Hiroyuki]

On Tue, 13 Nov 2007 13:38:11 -0800 (PST)
Christoph Lameter <clameter@sgi.com> wrote:
> Well that is currently not done for !SPARSEMEM configuration where 
> SECTIONS_WIDTH is also zero. So I left it as is.
> 
> > page_to_section is used in page_to_nid() if NODE_NOT_IN_PAGE_FLAGS=y.
> > (I'm not sure exact config dependency.)
> 
> NODE_NOT_IN_PAGE_FLAGS=y only occurs when flag bits are 
> taken away by sparsemem for the section bits.
> 
> 
Ahh, thank you for confirmation.

Regards,
-Kame

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>