From: "Martin J. Bligh" <mbligh@mbligh.org>
To: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: linux-mm <linux-mm@kvack.org>, Andy Whitcroft <apw@shadowen.org>
Subject: Re: [RFC][PATCH] remove zone_mem_map [3/4] pfn_to_page()
Date: Tue, 31 Jan 2006 21:07:18 -0800 [thread overview]
Message-ID: <43E04206.8030104@mbligh.org> (raw)
In-Reply-To: <43E02AAC.7050104@jp.fujitsu.com>
KAMEZAWA Hiroyuki wrote:
> Replace page_to_pfn() functions which uses zone->zone_mem_map.
>
> Although pfn_to_page() uses node->node_mem_map, page_to_pfn() uses
> zone->zone_mem_map. I don't know why. This patch make page_to_pfn
> use node->node_mem_map
I think that might have been because it's a faster op, at least on
some architectures, under sparsemem. Andy, can you confirm / deny?
Some part of the mapping was embedded in the page flags.
If Andy can't recall, I'll look again on Thursday when I'm back ...
M.
> x86_64 arch seems to want not to make page_to_pfn() inlined.
> So, CONFIG_DONT_INLINE_PAGE_TO_PFN is added.
> Other archs wants this ?
>
>
> Signed-Off-By: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
>
> Index: hogehoge/include/asm-i386/mmzone.h
> ===================================================================
> --- hogehoge.orig/include/asm-i386/mmzone.h
> +++ hogehoge/include/asm-i386/mmzone.h
> @@ -93,14 +93,6 @@ static inline int pfn_to_nid(unsigned lo
> &NODE_DATA(__node)->node_mem_map[node_localnr(__pfn,__node)]; \
> })
>
> -#define page_to_pfn(pg) \
> -({ \
> - struct page *__page = pg; \
> - struct zone *__zone = page_zone(__page); \
> - (unsigned long)(__page - __zone->zone_mem_map) \
> - + __zone->zone_start_pfn; \
> -})
> -
> #ifdef CONFIG_X86_NUMAQ /* we have contiguous memory on
> NUMA-Q */
> #define pfn_valid(pfn) ((pfn) < num_physpages)
> #else
> Index: hogehoge/include/linux/mm.h
> ===================================================================
> --- hogehoge.orig/include/linux/mm.h
> +++ hogehoge/include/linux/mm.h
> @@ -484,6 +484,17 @@ static inline struct pglist_data *page_n
> return NODE_DATA(page_to_nid);
> }
>
> +#if defined(CONFIG_FLAT_NODE_MEM_MAP) && !defined(HAVE_ARCH_PAGE_TO_PFN)
> +#ifndef CONFIG_DONT_INLINE_PAGE_TO_PFN /* looks x86_64 people wants
> this */
> +static inline unsigned long page_to_pfn(struct page *page)
> +{
> + return (page - page_node(page)->node_mem_map) +
> page_node(page)->node_start_pfn;
> +}
> +#else
> +extern unsigned long page_to_pfn(struct page *page);
> +#endif
> +#endif
> +
> static inline unsigned long page_to_section(struct page *page)
> {
> return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
> Index: hogehoge/include/asm-alpha/mmzone.h
> ===================================================================
> --- hogehoge.orig/include/asm-alpha/mmzone.h
> +++ hogehoge/include/asm-alpha/mmzone.h
> @@ -86,8 +86,7 @@ PLAT_NODE_DATA_LOCALNR(unsigned long p,
> pte_t pte; \
> unsigned long pfn; \
> \
> - pfn = ((unsigned long)((page)-page_zone(page)->zone_mem_map)) << 32; \
> - pfn += page_zone(page)->zone_start_pfn << 32; \
> + pfn = page_to_pfn(page) << 32; \
> pte_val(pte) = pfn | pgprot_val(pgprot); \
> \
> pte; \
> @@ -110,13 +109,9 @@ PLAT_NODE_DATA_LOCALNR(unsigned long p,
> (NODE_DATA(kvaddr_to_nid(kaddr))->node_mem_map +
> local_mapnr(kaddr)); \
> })
>
> -#define page_to_pfn(page) \
> - ((page) - page_zone(page)->zone_mem_map + \
> - (page_zone(page)->zone_start_pfn))
>
> #define page_to_pa(page) \
> - ((( (page) - page_zone(page)->zone_mem_map ) \
> - + page_zone(page)->zone_start_pfn) << PAGE_SHIFT)
> + (page_to_pfn(page) << PAGE_SHIFT)
>
> #define pfn_to_nid(pfn) pa_to_nid(((u64)(pfn) << PAGE_SHIFT))
> #define pfn_valid(pfn) \
> Index: hogehoge/include/asm-arm/memory.h
> ===================================================================
> --- hogehoge.orig/include/asm-arm/memory.h
> +++ hogehoge/include/asm-arm/memory.h
> @@ -190,9 +190,7 @@ static inline __deprecated void *bus_to_
> */
> #include <linux/numa.h>
>
> -#define page_to_pfn(page) \
> - (( (page) - page_zone(page)->zone_mem_map) \
> - + page_zone(page)->zone_start_pfn)
> +/* page_to_pfn is defined in include/linux/mm.h */
>
> #define pfn_to_page(pfn) \
> (PFN_TO_MAPBASE(pfn) + LOCAL_MAP_NR((pfn) << PAGE_SHIFT))
> Index: hogehoge/include/asm-m32r/mmzone.h
> ===================================================================
> --- hogehoge.orig/include/asm-m32r/mmzone.h
> +++ hogehoge/include/asm-m32r/mmzone.h
> @@ -28,13 +28,6 @@ extern struct pglist_data *node_data[];
> &NODE_DATA(__node)->node_mem_map[node_localnr(__pfn,__node)]; \
> })
>
> -#define page_to_pfn(pg) \
> -({ \
> - struct page *__page = pg; \
> - struct zone *__zone = page_zone(__page); \
> - (unsigned long)(__page - __zone->zone_mem_map) \
> - + __zone->zone_start_pfn; \
> -})
> #define pmd_page(pmd) (pfn_to_page(pmd_val(pmd) >> PAGE_SHIFT))
> /*
> * pfn_valid should be made as fast as possible, and the current
> definition
> Index: hogehoge/include/asm-mips/mmzone.h
> ===================================================================
> --- hogehoge.orig/include/asm-mips/mmzone.h
> +++ hogehoge/include/asm-mips/mmzone.h
> @@ -29,13 +29,6 @@
> __pg->node_mem_map + (__pfn - __pg->node_start_pfn); \
> })
>
> -#define page_to_pfn(p) \
> -({ \
> - struct page *__p = (p); \
> - struct zone *__z = page_zone(__p); \
> - ((__p - __z->zone_mem_map) + __z->zone_start_pfn); \
> -})
> -
> /* XXX: FIXME -- wli */
> #define kern_addr_valid(addr) (0)
>
> Index: hogehoge/include/asm-parisc/mmzone.h
> ===================================================================
> --- hogehoge.orig/include/asm-parisc/mmzone.h
> +++ hogehoge/include/asm-parisc/mmzone.h
> @@ -34,14 +34,6 @@ extern struct node_map_data node_data[];
> &NODE_DATA(__node)->node_mem_map[node_localnr(__pfn,__node)]; \
> })
>
> -#define page_to_pfn(pg) \
> -({ \
> - struct page *__page = pg; \
> - struct zone *__zone = page_zone(__page); \
> - BUG_ON(__zone == NULL); \
> - (unsigned long)(__page - __zone->zone_mem_map) \
> - + __zone->zone_start_pfn; \
> -})
>
> /* We have these possible memory map layouts:
> * Astro: 0-3.75, 67.75-68, 4-64
> Index: hogehoge/arch/x86_64/Kconfig
> ===================================================================
> --- hogehoge.orig/arch/x86_64/Kconfig
> +++ hogehoge/arch/x86_64/Kconfig
> @@ -321,6 +321,10 @@ config HAVE_ARCH_EARLY_PFN_TO_NID
> def_bool y
> depends on NUMA
>
> +config DONT_INLINE_PAGE_TO_PFN
> + def_bool y
> + depends on NUMA
> +
> config NR_CPUS
> int "Maximum number of CPUs (2-256)"
> range 2 256
> Index: hogehoge/arch/x86_64/mm/numa.c
> ===================================================================
> --- hogehoge.orig/arch/x86_64/mm/numa.c
> +++ hogehoge/arch/x86_64/mm/numa.c
> @@ -379,8 +379,8 @@ EXPORT_SYMBOL(pfn_to_page);
>
> unsigned long page_to_pfn(struct page *page)
> {
> - return (long)(((page) - page_zone(page)->zone_mem_map) +
> - page_zone(page)->zone_start_pfn);
> + return (long)(((page) - page_node(page)->node_mem_map) +
> + page_node(page)->node_start_pfn);
> }
> EXPORT_SYMBOL(page_to_pfn);
>
> Index: hogehoge/mm/page_alloc.c
> ===================================================================
> --- hogehoge.orig/mm/page_alloc.c
> +++ hogehoge/mm/page_alloc.c
> @@ -2712,3 +2712,13 @@ void *__init alloc_large_system_hash(con
>
> return table;
> }
> +
> +/* inlined version is in include/linux/mm.h */
> +#if defined(CONFIG_NODE_FLAT_MEM_MAP) && defined(CONFIG_DISCONTIG_MEM)
> +#ifdef DONT_INLINE_PAGE_TO_PFN
> +unsigned long page_to_pfn(struct page *page)
> +{
> + return (page - page_node(page)->node_mem_map) +
> page_node(page)->node_start_pfn;
> +}
> +#endif
> +#endif
> Index: hogehoge/arch/ia64/Kconfig
> ===================================================================
> --- hogehoge.orig/arch/ia64/Kconfig
> +++ hogehoge/arch/ia64/Kconfig
> @@ -330,6 +330,10 @@ config VIRTUAL_MEM_MAP
> require the DISCONTIGMEM option for your machine. If you are
> unsure, say Y.
>
> +config HAVE_ARCH_PAGE_TO_PFN
> + default y
> + if VIRTUAL_MEM_MAP
> +
> config HOLES_IN_ZONE
> bool
> default y if VIRTUAL_MEM_MAP
> Index: hogehoge/mm/Kconfig
> ===================================================================
> --- hogehoge.orig/mm/Kconfig
> +++ hogehoge/mm/Kconfig
> @@ -77,6 +77,10 @@ config FLAT_NODE_MEM_MAP
> def_bool y
> depends on !SPARSEMEM
>
> +config HAVE_ARCH_PAGE_TO_PFN
> + def_bool y
> + depends on FLATMEM
> +
> #
> # Both the NUMA code and DISCONTIGMEM use arrays of pg_data_t's
> # to represent different areas of memory. This variable allows
>
>
--
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>
next prev parent reply other threads:[~2006-02-01 5:07 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2006-02-01 3:27 KAMEZAWA Hiroyuki
2006-02-01 5:07 ` Martin J. Bligh [this message]
2006-02-01 5:21 ` KAMEZAWA Hiroyuki
2006-02-01 13:34 ` Andy Whitcroft
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=43E04206.8030104@mbligh.org \
--to=mbligh@mbligh.org \
--cc=apw@shadowen.org \
--cc=kamezawa.hiroyu@jp.fujitsu.com \
--cc=linux-mm@kvack.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox