thp: keep huge zero page pinned until tlb flush

Andrea has found[1] a race condition on MMU-gather based TLB flush vs
split_huge_page() or shrinker which frees huge zero under us (patch 1/2
and 2/2 respectively).

With new THP refcounting, we don't need patch 1/2: mmu_gather keeps the
page pinned until flush is complete and the pin prevents the page from
being split under us.

We still need patch 2/2.  This is simplified version of Andrea's patch.
We don't need fancy encoding.

[1] http://lkml.kernel.org/r/[email protected]

Signed-off-by: Kirill A. Shutemov <[email protected]>
Reported-by: Andrea Arcangeli <[email protected]>
Reviewed-by: Andrea Arcangeli <[email protected]>
Cc: "Aneesh Kumar K.V" <[email protected]>
Cc: Mel Gorman <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: Vlastimil Babka <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

include/linux/huge_mm.h

@@ -152,6 +152,7 @@ static inline bool is_huge_zero_pmd(pmd_t pmd)
 }
 
 struct page *get_huge_zero_page(void);
+void put_huge_zero_page(void);
 
 #else /* CONFIG_TRANSPARENT_HUGEPAGE */
 #define HPAGE_PMD_SHIFT ({ BUILD_BUG(); 0; })
@@ -208,6 +209,10 @@ static inline bool is_huge_zero_page(struct page *page)
 	return false;
 }
 
+static inline void put_huge_zero_page(void)
+{
+	BUILD_BUG();
+}
 
 static inline struct page *follow_devmap_pmd(struct vm_area_struct *vma,
 		unsigned long addr, pmd_t *pmd, int flags)

mm/huge_memory.c

@@ -232,7 +232,7 @@ struct page *get_huge_zero_page(void)
 	return READ_ONCE(huge_zero_page);
 }
 
-static void put_huge_zero_page(void)
+void put_huge_zero_page(void)
 {
 	/*
 	 * Counter should never go to zero here. Only shrinker can put
@@ -1684,12 +1684,12 @@ int zap_huge_pmd(struct mmu_gather *tlb, struct vm_area_struct *vma,
 	if (vma_is_dax(vma)) {
 		spin_unlock(ptl);
 		if (is_huge_zero_pmd(orig_pmd))
-			put_huge_zero_page();
+			tlb_remove_page(tlb, pmd_page(orig_pmd));
 	} else if (is_huge_zero_pmd(orig_pmd)) {
 		pte_free(tlb->mm, pgtable_trans_huge_withdraw(tlb->mm, pmd));
 		atomic_long_dec(&tlb->mm->nr_ptes);
 		spin_unlock(ptl);
-		put_huge_zero_page();
+		tlb_remove_page(tlb, pmd_page(orig_pmd));
 	} else {
 		struct page *page = pmd_page(orig_pmd);
 		page_remove_rmap(page, true);

mm/swap.c

@@ -728,6 +728,11 @@ void release_pages(struct page **pages, int nr, bool cold)
 			zone = NULL;
 		}
 
+		if (is_huge_zero_page(page)) {
+			put_huge_zero_page();
+			continue;
+		}
+
 		page = compound_head(page);
 		if (!put_page_testzero(page))
 			continue;