[RFC PATCH 06/14] mm/rmap: avoid flushing on page_vma_mkclean_one() when possible

[Nadav Amit <nadav.amit@gmail.com>]

From: Nadav Amit <namit@vmware.com>

x86 is capable to avoid TLB flush on clean writable entries.
page_vma_mkclean_one() does not take advantage of this behavior. Adapt
it.

Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Andrew Cooper <andrew.cooper3@citrix.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Will Deacon <will@kernel.org>
Cc: Yu Zhao <yuzhao@google.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Nadav Amit <namit@vmware.com>
---
 mm/rmap.c | 16 ++++++++++++----
 1 file changed, 12 insertions(+), 4 deletions(-)

diff --git a/mm/rmap.c b/mm/rmap.c
index 83172ee0ea35..23997c387858 100644
--- a/mm/rmap.c
+++ b/mm/rmap.c
@@ -961,17 +961,25 @@ static int page_vma_mkclean_one(struct page_vma_mapped_walk *pvmw)
 
 		address = pvmw->address;
 		if (pvmw->pte) {
-			pte_t entry;
+			pte_t entry, oldpte;
 			pte_t *pte = pvmw->pte;
 
 			if (!pte_dirty(*pte) && !pte_write(*pte))
 				continue;
 
 			flush_cache_page(vma, address, pte_pfn(*pte));
-			entry = ptep_clear_flush(vma, address, pte);
-			entry = pte_wrprotect(entry);
+			oldpte = ptep_modify_prot_start(pvmw->vma, address,
+							pte);
+
+			entry = pte_wrprotect(oldpte);
 			entry = pte_mkclean(entry);
-			set_pte_at(vma->vm_mm, address, pte, entry);
+
+			if (pte_needs_flush(oldpte, entry) ||
+			    mm_tlb_flush_pending(vma->vm_mm))
+				flush_tlb_page(vma, address);
+
+			ptep_modify_prot_commit(vma, address, pte, oldpte,
+						entry);
 			ret = 1;
 		} else {
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
-- 
2.25.1