mm: No need to check pcp page when free it to buddy
Subsystems (2)
PATCH v2 1/1 mm: No need to check pcp page when free it to buddy
From: Zhaoyang Huang <[email protected]>
The pages on pcp list have been checked the validity via bellowing call chain. It is no need to check it again when free them back to buddy by free_pcppages_bulk.
free_unref_page\free_unref_page_list
{
...
//page's validity is checked here if (!free_unref_page_prepare(page, pfn, order)) return; ... //page added to pcp->lists and status is ready for moving to buddy list_add(&page->lru, &pcp->lists[pindex]);
---
v2: update code and comment for clarifying
---
--- mm/page_alloc.c | 16 ---------------- 1 file changed, 16 deletions(-)
mm/page_alloc.c +0 -16
@@ -1420,14 +1420,6 @@ static bool free_pcp_prepare(struct page *page, unsigned int order) {return free_pages_prepare (page, order, true,FPI_NONE ); } - -static bool bulkfree_pcp_prepare (struct page *page) -{ -if (debug_pagealloc_enabled_static ()) -return check_free_page (page); -else -return false; -}#else /* * With DEBUG_VM disabled, order-0 pages being freed are checked only when
@@ -1442,11 +1434,6 @@ static bool free_pcp_prepare(struct page *page, unsigned int order) else return free_pages_prepare(page, order, false, FPI_NONE); } - -static bool bulkfree_pcp_prepare (struct page *page) -{ -return check_free_page (page); -}#endif /* CONFIG_DEBUG_VM */ /*
@@ -1512,9 +1499,6 @@ static void free_pcppages_bulk(struct zone *zone, int count, count -= nr_pages; pcp->count -= nr_pages; -if (bulkfree_pcp_prepare (page)) -continue ; -/* MIGRATE_ISOLATE page should not go to pcplists */ VM_BUG_ON_PAGE (is_migrate_isolate (mt), page);/* Pageblock could have been isolated meanwhile */
Signature · 1 line
1.9.1
Andrew Morton
maintainerQuoted 8 lines · On Thu, 15 Sep 2022 10:40:54 +0800 "zhaoyang.huang"…
From: Zhaoyang Huang <[email protected]>
The pages on pcp list have been checked the validity via bellowing call chain. It is no need to check it again when free them back to buddy by
free_pcppages_bulk.free_unref_page\free_unref_page_list
{
...
Quoted 7 lines
//page's validity is checked here if (!free_unref_page_prepare(page, pfn, order)) return; ... //page added to pcp->lists and status is ready for moving to buddy list_add(&page->lru, &pcp->lists[pindex]);
The comment over free_pcp_prepare():
/*
* With DEBUG_VM enabled, order-0 pages are checked immediately when being freed
* to pcp lists. With debug_pagealloc also enabled, they are also rechecked when
* moved from pcp lists to free lists.
*/
Which seems sensible - a page's state may have been messed up while it's in the pcp lists.