From: Mel Gorman Date: Wed, 20 Feb 2013 02:14:43 +0000 (+1100) Subject: mm: uninline page_xchg_last_nid() X-Git-Tag: next-20130220~1^2~451 X-Git-Url: https://git.karo-electronics.de/?a=commitdiff_plain;h=1b7513bef804c84da9e2e3db2c97b4c11d38d401;p=karo-tx-linux.git mm: uninline page_xchg_last_nid() Andrew Morton pointed out that page_xchg_last_nid() and reset_page_last_nid() were "getting nuttily large" and asked that it be investigated. reset_page_last_nid() is on the page free path and it would be unfortunate to make that path more expensive than it needs to be. Due to the internal use of page_xchg_last_nid() it is already too expensive but fortunately, it should also be impossible for the page->flags to be updated in parallel when we call reset_page_last_nid(). Instead of unlining the function, it uses a simplier implementation that assumes no parallel updates and should now be sufficiently short for inlining. page_xchg_last_nid() is called in paths that are already quite expensive (splitting huge page, fault handling, migration) and it is reasonable to uninline. There was not really a good place to place the function but mm/mmzone.c was the closest fit IMO. This patch saved 128 bytes of text in the vmlinux file for the kernel configuration I used for testing automatic NUMA balancing. Signed-off-by: Mel Gorman Signed-off-by: Andrew Morton --- diff --git a/include/linux/mm.h b/include/linux/mm.h index c9759ea90dad..0b5e2a5f366b 100644 --- a/include/linux/mm.h +++ b/include/linux/mm.h @@ -679,25 +679,14 @@ static inline int page_last_nid(struct page *page) return (page->flags >> LAST_NID_PGSHIFT) & LAST_NID_MASK; } -static inline int page_xchg_last_nid(struct page *page, int nid) -{ - unsigned long old_flags, flags; - int last_nid; - - do { - old_flags = flags = page->flags; - last_nid = page_last_nid(page); - - flags &= ~(LAST_NID_MASK << LAST_NID_PGSHIFT); - flags |= (nid & LAST_NID_MASK) << LAST_NID_PGSHIFT; - } while (unlikely(cmpxchg(&page->flags, old_flags, flags) != old_flags)); - - return last_nid; -} +extern int page_xchg_last_nid(struct page *page, int nid); static inline void reset_page_last_nid(struct page *page) { - page_xchg_last_nid(page, (1 << LAST_NID_SHIFT) - 1); + int nid = (1 << LAST_NID_SHIFT) - 1; + + page->flags &= ~(LAST_NID_MASK << LAST_NID_PGSHIFT); + page->flags |= (nid & LAST_NID_MASK) << LAST_NID_PGSHIFT; } #endif /* LAST_NID_NOT_IN_PAGE_FLAGS */ #else diff --git a/mm/mmzone.c b/mm/mmzone.c index 4596d81b89b1..bce796e8487f 100644 --- a/mm/mmzone.c +++ b/mm/mmzone.c @@ -1,7 +1,7 @@ /* * linux/mm/mmzone.c * - * management codes for pgdats and zones. + * management codes for pgdats, zones and page flags */ @@ -96,3 +96,21 @@ void lruvec_init(struct lruvec *lruvec) for_each_lru(lru) INIT_LIST_HEAD(&lruvec->lists[lru]); } + +#if defined(CONFIG_NUMA_BALANCING) && !defined(LAST_NID_NOT_IN_PAGE_FLAGS) +int page_xchg_last_nid(struct page *page, int nid) +{ + unsigned long old_flags, flags; + int last_nid; + + do { + old_flags = flags = page->flags; + last_nid = page_last_nid(page); + + flags &= ~(LAST_NID_MASK << LAST_NID_PGSHIFT); + flags |= (nid & LAST_NID_MASK) << LAST_NID_PGSHIFT; + } while (unlikely(cmpxchg(&page->flags, old_flags, flags) != old_flags)); + + return last_nid; +} +#endif