[Pmem] [PATCH 05/15] NUMA balancing: migrate dirty file cache pages

From: Huang Ying <ying.huang(a)intel.com> ANBZ: #80 cherry-picked from https://git.kernel.org/pub/scm/linux/kernel/git/vishal/tiering.git/commit/?… Commonly used file systems such as ext4 and xfs can migrate dirty file cache pages with MIGRATE_ASYNC. So enable the dirty file cache pages migration in AutoNUMA to optimize the page placement for these pages. The patch improves the read/write bandwidth of fio benchmark with 80:20 read/write ratio and normal random distribution by 11.3% on a 2 socket Intel server with Optane DC Persistent Memory Model. Signed-off-by: "Huang, Ying" <ying.huang(a)intel.com> Signed-off-by: Baolin Wang <baolin.wang(a)linux.alibaba.com> --- mm/migrate.c | 7 ------- mm/mprotect.c | 8 -------- 2 files changed, 15 deletions(-) diff --git a/mm/migrate.c b/mm/migrate.c index c7632c9..4a57837 100644 --- a/mm/migrate.c +++ b/mm/migrate.c @@ -2060,13 +2060,6 @@ int migrate_misplaced_page(struct page *page, struct vm_area_struct *vma, (vma->vm_flags & VM_EXEC)) goto out; - /* - * Also do not migrate dirty pages as not all filesystems can move - * dirty pages in MIGRATE_ASYNC mode which is a waste of cycles. - */ - if (page_is_file_cache(page) && PageDirty(page)) - goto out; - isolated = numamigrate_isolate_page(pgdat, page); if (!isolated) goto out; diff --git a/mm/mprotect.c b/mm/mprotect.c index 3072213..6525e96 100644 --- a/mm/mprotect.c +++ b/mm/mprotect.c @@ -93,14 +93,6 @@ static unsigned long change_pte_range(struct vm_area_struct *vma, pmd_t *pmd, page_mapcount(page) != 1) continue; - /* - * While migration can move some dirty pages, - * it cannot move them all from MIGRATE_ASYNC - * context. - */ - if (page_is_file_cache(page) && PageDirty(page)) - continue; - /* Avoid TLB flush if possible */ if (pte_protnone(oldpte)) continue; -- 1.8.3.1