1 files changed, 177 insertions, 85 deletions

diff --git a/mm/khugepaged.c b/mm/khugepaged.c
index cd280afb246e..3f032487825b 100644
--- a/mm/khugepaged.c
+++ b/mm/khugepaged.c
@@ -28,6 +28,8 @@ enum scan_result {
 	SCAN_SUCCEED,
 	SCAN_PMD_NULL,
 	SCAN_EXCEED_NONE_PTE,
+	SCAN_EXCEED_SWAP_PTE,
+	SCAN_EXCEED_SHARED_PTE,
 	SCAN_PTE_NON_PRESENT,
 	SCAN_PTE_UFFD_WP,
 	SCAN_PAGE_RO,
@@ -47,7 +49,6 @@ enum scan_result {
 	SCAN_DEL_PAGE_LRU,
 	SCAN_ALLOC_HUGE_PAGE_FAIL,
 	SCAN_CGROUP_CHARGE_FAIL,
-	SCAN_EXCEED_SWAP_PTE,
 	SCAN_TRUNCATED,
 	SCAN_PAGE_HAS_PRIVATE,
 };
@@ -72,6 +73,7 @@ static DECLARE_WAIT_QUEUE_HEAD(khugepaged_wait);
  */
 static unsigned int khugepaged_max_ptes_none __read_mostly;
 static unsigned int khugepaged_max_ptes_swap __read_mostly;
+static unsigned int khugepaged_max_ptes_shared __read_mostly;
 
 #define MM_SLOTS_HASH_BITS 10
 static __read_mostly DEFINE_HASHTABLE(mm_slots_hash, MM_SLOTS_HASH_BITS);
@@ -291,15 +293,43 @@ static struct kobj_attribute khugepaged_max_ptes_swap_attr =
 	__ATTR(max_ptes_swap, 0644, khugepaged_max_ptes_swap_show,
 	       khugepaged_max_ptes_swap_store);
 
+static ssize_t khugepaged_max_ptes_shared_show(struct kobject *kobj,
+					     struct kobj_attribute *attr,
+					     char *buf)
+{
+	return sprintf(buf, "%u\n", khugepaged_max_ptes_shared);
+}
+
+static ssize_t khugepaged_max_ptes_shared_store(struct kobject *kobj,
+					      struct kobj_attribute *attr,
+					      const char *buf, size_t count)
+{
+	int err;
+	unsigned long max_ptes_shared;
+
+	err  = kstrtoul(buf, 10, &max_ptes_shared);
+	if (err || max_ptes_shared > HPAGE_PMD_NR-1)
+		return -EINVAL;
+
+	khugepaged_max_ptes_shared = max_ptes_shared;
+
+	return count;
+}
+
+static struct kobj_attribute khugepaged_max_ptes_shared_attr =
+	__ATTR(max_ptes_shared, 0644, khugepaged_max_ptes_shared_show,
+	       khugepaged_max_ptes_shared_store);
+
 static struct attribute *khugepaged_attr[] = {
 	&khugepaged_defrag_attr.attr,
 	&khugepaged_max_ptes_none_attr.attr,
+	&khugepaged_max_ptes_swap_attr.attr,
+	&khugepaged_max_ptes_shared_attr.attr,
 	&pages_to_scan_attr.attr,
 	&pages_collapsed_attr.attr,
 	&full_scans_attr.attr,
 	&scan_sleep_millisecs_attr.attr,
 	&alloc_sleep_millisecs_attr.attr,
-	&khugepaged_max_ptes_swap_attr.attr,
 	NULL,
 };
 
@@ -359,6 +389,7 @@ int __init khugepaged_init(void)
 	khugepaged_pages_to_scan = HPAGE_PMD_NR * 8;
 	khugepaged_max_ptes_none = HPAGE_PMD_NR - 1;
 	khugepaged_max_ptes_swap = HPAGE_PMD_NR / 8;
+	khugepaged_max_ptes_shared = HPAGE_PMD_NR / 2;
 
 	return 0;
 }
@@ -512,27 +543,52 @@ void __khugepaged_exit(struct mm_struct *mm)
 
 static void release_pte_page(struct page *page)
 {
-	dec_node_page_state(page, NR_ISOLATED_ANON + page_is_file_lru(page));
+	mod_node_page_state(page_pgdat(page),
+			NR_ISOLATED_ANON + page_is_file_lru(page),
+			-compound_nr(page));
 	unlock_page(page);
 	putback_lru_page(page);
 }
 
-static void release_pte_pages(pte_t *pte, pte_t *_pte)
+static void release_pte_pages(pte_t *pte, pte_t *_pte,
+		struct list_head *compound_pagelist)
 {
+	struct page *page, *tmp;
+
 	while (--_pte >= pte) {
 		pte_t pteval = *_pte;
-		if (!pte_none(pteval) && !is_zero_pfn(pte_pfn(pteval)))
-			release_pte_page(pte_page(pteval));
+
+		page = pte_page(pteval);
+		if (!pte_none(pteval) && !is_zero_pfn(pte_pfn(pteval)) &&
+				!PageCompound(page))
+			release_pte_page(page);
+	}
+
+	list_for_each_entry_safe(page, tmp, compound_pagelist, lru) {
+		list_del(&page->lru);
+		release_pte_page(page);
 	}
 }
 
+static bool is_refcount_suitable(struct page *page)
+{
+	int expected_refcount;
+
+	expected_refcount = total_mapcount(page);
+	if (PageSwapCache(page))
+		expected_refcount += compound_nr(page);
+
+	return page_count(page) == expected_refcount;
+}
+
 static int __collapse_huge_page_isolate(struct vm_area_struct *vma,
 					unsigned long address,
-					pte_t *pte)
+					pte_t *pte,
+					struct list_head *compound_pagelist)
 {
 	struct page *page = NULL;
 	pte_t *_pte;
-	int none_or_zero = 0, result = 0, referenced = 0;
+	int none_or_zero = 0, shared = 0, result = 0, referenced = 0;
 	bool writable = false;
 
 	for (_pte = pte; _pte < pte+HPAGE_PMD_NR;
@@ -558,13 +614,27 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma,
 			goto out;
 		}
 
-		/* TODO: teach khugepaged to collapse THP mapped with pte */
-		if (PageCompound(page)) {
-			result = SCAN_PAGE_COMPOUND;
+		VM_BUG_ON_PAGE(!PageAnon(page), page);
+
+		if (page_mapcount(page) > 1 &&
+				++shared > khugepaged_max_ptes_shared) {
+			result = SCAN_EXCEED_SHARED_PTE;
 			goto out;
 		}
 
-		VM_BUG_ON_PAGE(!PageAnon(page), page);
+		if (PageCompound(page)) {
+			struct page *p;
+			page = compound_head(page);
+
+			/*
+			 * Check if we have dealt with the compound page
+			 * already
+			 */
+			list_for_each_entry(p, compound_pagelist, lru) {
+				if (page == p)
+					goto next;
+			}
+		}
 
 		/*
 		 * We can do it before isolate_lru_page because the
@@ -578,28 +648,30 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma,
 		}
 
 		/*
-		 * cannot use mapcount: can't collapse if there's a gup pin.
-		 * The page must only be referenced by the scanned process
-		 * and page swap cache.
+		 * Check if the page has any GUP (or other external) pins.
+		 *
+		 * The page table that maps the page has been already unlinked
+		 * from the page table tree and this process cannot get
+		 * an additinal pin on the page.
+		 *
+		 * New pins can come later if the page is shared across fork,
+		 * but not from this process. The other process cannot write to
+		 * the page, only trigger CoW.
 		 */
-		if (page_count(page) != 1 + PageSwapCache(page)) {
+		if (!is_refcount_suitable(page)) {
 			unlock_page(page);
 			result = SCAN_PAGE_COUNT;
 			goto out;
 		}
-		if (pte_write(pteval)) {
-			writable = true;
-		} else {
-			if (PageSwapCache(page) &&
-			    !reuse_swap_page(page, NULL)) {
-				unlock_page(page);
-				result = SCAN_SWAP_CACHE_PAGE;
-				goto out;
-			}
+		if (!pte_write(pteval) && PageSwapCache(page) &&
+				!reuse_swap_page(page, NULL)) {
 			/*
-			 * Page is not in the swap cache. It can be collapsed
-			 * into a THP.
+			 * Page is in the swap cache and cannot be re-used.
+			 * It cannot be collapsed into a THP.
 			 */
+			unlock_page(page);
+			result = SCAN_SWAP_CACHE_PAGE;
+			goto out;
 		}
 
 		/*
@@ -611,16 +683,23 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma,
 			result = SCAN_DEL_PAGE_LRU;
 			goto out;
 		}
-		inc_node_page_state(page,
-				NR_ISOLATED_ANON + page_is_file_lru(page));
+		mod_node_page_state(page_pgdat(page),
+				NR_ISOLATED_ANON + page_is_file_lru(page),
+				compound_nr(page));
 		VM_BUG_ON_PAGE(!PageLocked(page), page);
 		VM_BUG_ON_PAGE(PageLRU(page), page);
 
+		if (PageCompound(page))
+			list_add_tail(&page->lru, compound_pagelist);
+next:
 		/* There should be enough young pte to collapse the page */
 		if (pte_young(pteval) ||
 		    page_is_young(page) || PageReferenced(page) ||
 		    mmu_notifier_test_young(vma->vm_mm, address))
 			referenced++;
+
+		if (pte_write(pteval))
+			writable = true;
 	}
 	if (likely(writable)) {
 		if (likely(referenced)) {
@@ -634,7 +713,7 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma,
 	}
 
 out:
-	release_pte_pages(pte, _pte);
+	release_pte_pages(pte, _pte, compound_pagelist);
 	trace_mm_collapse_huge_page_isolate(page, none_or_zero,
 					    referenced, writable, result);
 	return 0;
@@ -643,13 +722,14 @@ out:
 static void __collapse_huge_page_copy(pte_t *pte, struct page *page,
 				      struct vm_area_struct *vma,
 				      unsigned long address,
-				      spinlock_t *ptl)
+				      spinlock_t *ptl,
+				      struct list_head *compound_pagelist)
 {
+	struct page *src_page, *tmp;
 	pte_t *_pte;
 	for (_pte = pte; _pte < pte + HPAGE_PMD_NR;
 				_pte++, page++, address += PAGE_SIZE) {
 		pte_t pteval = *_pte;
-		struct page *src_page;
 
 		if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) {
 			clear_user_highpage(page, address);
@@ -669,8 +749,8 @@ static void __collapse_huge_page_copy(pte_t *pte, struct page *page,
 		} else {
 			src_page = pte_page(pteval);
 			copy_user_highpage(page, src_page, address, vma);
-			VM_BUG_ON_PAGE(page_mapcount(src_page) != 1, src_page);
-			release_pte_page(src_page);
+			if (!PageCompound(src_page))
+				release_pte_page(src_page);
 			/*
 			 * ptl mostly unnecessary, but preempt has to
 			 * be disabled to update the per-cpu stats
@@ -687,6 +767,11 @@ static void __collapse_huge_page_copy(pte_t *pte, struct page *page,
 			free_page_and_swap_cache(src_page);
 		}
 	}
+
+	list_for_each_entry_safe(src_page, tmp, compound_pagelist, lru) {
+		list_del(&src_page->lru);
+		release_pte_page(src_page);
+	}
 }
 
 static void khugepaged_alloc_sleep(void)
@@ -899,11 +984,6 @@ static bool __collapse_huge_page_swapin(struct mm_struct *mm,
 		.pgoff = linear_page_index(vma, address),
 	};
 
-	/* we only decide to swapin, if there is enough young ptes */
-	if (referenced < HPAGE_PMD_NR/2) {
-		trace_mm_collapse_huge_page_swapin(mm, swapped_in, referenced, 0);
-		return false;
-	}
 	vmf.pte = pte_offset_map(pmd, address);
 	for (; vmf.address < address + HPAGE_PMD_NR*PAGE_SIZE;
 			vmf.pte++, vmf.address += PAGE_SIZE) {
@@ -936,6 +1016,11 @@ static bool __collapse_huge_page_swapin(struct mm_struct *mm,
 	}
 	vmf.pte--;
 	pte_unmap(vmf.pte);
+
+	/* Drain LRU add pagevec to remove extra pin on the swapped in pages */
+	if (swapped_in)
+		lru_add_drain();
+
 	trace_mm_collapse_huge_page_swapin(mm, swapped_in, referenced, 1);
 	return true;
 }
@@ -943,15 +1028,15 @@ static bool __collapse_huge_page_swapin(struct mm_struct *mm,
 static void collapse_huge_page(struct mm_struct *mm,
 				   unsigned long address,
 				   struct page **hpage,
-				   int node, int referenced)
+				   int node, int referenced, int unmapped)
 {
+	LIST_HEAD(compound_pagelist);
 	pmd_t *pmd, _pmd;
 	pte_t *pte;
 	pgtable_t pgtable;
 	struct page *new_page;
 	spinlock_t *pmd_ptl, *pte_ptl;
 	int isolated = 0, result = 0;
-	struct mem_cgroup *memcg;
 	struct vm_area_struct *vma;
 	struct mmu_notifier_range range;
 	gfp_t gfp;
@@ -974,15 +1059,15 @@ static void collapse_huge_page(struct mm_struct *mm,
 		goto out_nolock;
 	}
 
-	if (unlikely(mem_cgroup_try_charge(new_page, mm, gfp, &memcg, true))) {
+	if (unlikely(mem_cgroup_charge(new_page, mm, gfp))) {
 		result = SCAN_CGROUP_CHARGE_FAIL;
 		goto out_nolock;
 	}
+	count_memcg_page_event(new_page, THP_COLLAPSE_ALLOC);
 
 	down_read(&mm->mmap_sem);
 	result = hugepage_vma_revalidate(mm, address, &vma);
 	if (result) {
-		mem_cgroup_cancel_charge(new_page, memcg, true);
 		up_read(&mm->mmap_sem);
 		goto out_nolock;
 	}
@@ -990,7 +1075,6 @@ static void collapse_huge_page(struct mm_struct *mm,
 	pmd = mm_find_pmd(mm, address);
 	if (!pmd) {
 		result = SCAN_PMD_NULL;
-		mem_cgroup_cancel_charge(new_page, memcg, true);
 		up_read(&mm->mmap_sem);
 		goto out_nolock;
 	}
@@ -1000,8 +1084,8 @@ static void collapse_huge_page(struct mm_struct *mm,
 	 * If it fails, we release mmap_sem and jump out_nolock.
 	 * Continuing to collapse causes inconsistency.
 	 */
-	if (!__collapse_huge_page_swapin(mm, vma, address, pmd, referenced)) {
-		mem_cgroup_cancel_charge(new_page, memcg, true);
+	if (unmapped && !__collapse_huge_page_swapin(mm, vma, address,
+						     pmd, referenced)) {
 		up_read(&mm->mmap_sem);
 		goto out_nolock;
 	}
@@ -1044,7 +1128,8 @@ static void collapse_huge_page(struct mm_struct *mm,
 	mmu_notifier_invalidate_range_end(&range);
 
 	spin_lock(pte_ptl);
-	isolated = __collapse_huge_page_isolate(vma, address, pte);
+	isolated = __collapse_huge_page_isolate(vma, address, pte,
+			&compound_pagelist);
 	spin_unlock(pte_ptl);
 
 	if (unlikely(!isolated)) {
@@ -1069,7 +1154,8 @@ static void collapse_huge_page(struct mm_struct *mm,
 	 */
 	anon_vma_unlock_write(vma->anon_vma);
 
-	__collapse_huge_page_copy(pte, new_page, vma, address, pte_ptl);
+	__collapse_huge_page_copy(pte, new_page, vma, address, pte_ptl,
+			&compound_pagelist);
 	pte_unmap(pte);
 	__SetPageUptodate(new_page);
 	pgtable = pmd_pgtable(_pmd);
@@ -1087,8 +1173,6 @@ static void collapse_huge_page(struct mm_struct *mm,
 	spin_lock(pmd_ptl);
 	BUG_ON(!pmd_none(*pmd));
 	page_add_new_anon_rmap(new_page, vma, address, true);
-	mem_cgroup_commit_charge(new_page, memcg, false, true);
-	count_memcg_events(memcg, THP_COLLAPSE_ALLOC, 1);
 	lru_cache_add_active_or_unevictable(new_page, vma);
 	pgtable_trans_huge_deposit(mm, pmd, pgtable);
 	set_pmd_at(mm, address, pmd, _pmd);
@@ -1102,10 +1186,11 @@ static void collapse_huge_page(struct mm_struct *mm,
 out_up_write:
 	up_write(&mm->mmap_sem);
 out_nolock:
+	if (!IS_ERR_OR_NULL(*hpage))
+		mem_cgroup_uncharge(*hpage);
 	trace_mm_collapse_huge_page(mm, isolated, result);
 	return;
 out:
-	mem_cgroup_cancel_charge(new_page, memcg, true);
 	goto out_up_write;
 }
 
@@ -1116,7 +1201,8 @@ static int khugepaged_scan_pmd(struct mm_struct *mm,
 {
 	pmd_t *pmd;
 	pte_t *pte, *_pte;
-	int ret = 0, none_or_zero = 0, result = 0, referenced = 0;
+	int ret = 0, result = 0, referenced = 0;
+	int none_or_zero = 0, shared = 0;
 	struct page *page = NULL;
 	unsigned long _address;
 	spinlock_t *ptl;
@@ -1188,12 +1274,14 @@ static int khugepaged_scan_pmd(struct mm_struct *mm,
 			goto out_unmap;
 		}
 
-		/* TODO: teach khugepaged to collapse THP mapped with pte */
-		if (PageCompound(page)) {
-			result = SCAN_PAGE_COMPOUND;
+		if (page_mapcount(page) > 1 &&
+				++shared > khugepaged_max_ptes_shared) {
+			result = SCAN_EXCEED_SHARED_PTE;
 			goto out_unmap;
 		}
 
+		page = compound_head(page);
+
 		/*
 		 * Record which node the original page is from and save this
 		 * information to khugepaged_node_load[].
@@ -1220,11 +1308,23 @@ static int khugepaged_scan_pmd(struct mm_struct *mm,
 		}
 
 		/*
-		 * cannot use mapcount: can't collapse if there's a gup pin.
-		 * The page must only be referenced by the scanned process
-		 * and page swap cache.
+		 * Check if the page has any GUP (or other external) pins.
+		 *
+		 * Here the check is racy it may see totmal_mapcount > refcount
+		 * in some cases.
+		 * For example, one process with one forked child process.
+		 * The parent has the PMD split due to MADV_DONTNEED, then
+		 * the child is trying unmap the whole PMD, but khugepaged
+		 * may be scanning the parent between the child has
+		 * PageDoubleMap flag cleared and dec the mapcount.  So
+		 * khugepaged may see total_mapcount > refcount.
+		 *
+		 * But such case is ephemeral we could always retry collapse
+		 * later.  However it may report false positive if the page
+		 * has excessive GUP pins (i.e. 512).  Anyway the same check
+		 * will be done again later the risk seems low.
 		 */
-		if (page_count(page) != 1 + PageSwapCache(page)) {
+		if (!is_refcount_suitable(page)) {
 			result = SCAN_PAGE_COUNT;
 			goto out_unmap;
 		}
@@ -1233,22 +1333,21 @@ static int khugepaged_scan_pmd(struct mm_struct *mm,
 		    mmu_notifier_test_young(vma->vm_mm, address))
 			referenced++;
 	}
-	if (writable) {
-		if (referenced) {
-			result = SCAN_SUCCEED;
-			ret = 1;
-		} else {
-			result = SCAN_LACK_REFERENCED_PAGE;
-		}
-	} else {
+	if (!writable) {
 		result = SCAN_PAGE_RO;
+	} else if (!referenced || (unmapped && referenced < HPAGE_PMD_NR/2)) {
+		result = SCAN_LACK_REFERENCED_PAGE;
+	} else {
+		result = SCAN_SUCCEED;
+		ret = 1;
 	}
 out_unmap:
 	pte_unmap_unlock(pte, ptl);
 	if (ret) {
 		node = khugepaged_find_target_node();
 		/* collapse_huge_page will return with the mmap_sem released */
-		collapse_huge_page(mm, address, hpage, node, referenced);
+		collapse_huge_page(mm, address, hpage, node,
+				referenced, unmapped);
 	}
 out:
 	trace_mm_khugepaged_scan_pmd(mm, page, writable, referenced,
@@ -1515,7 +1614,6 @@ static void collapse_file(struct mm_struct *mm,
 	struct address_space *mapping = file->f_mapping;
 	gfp_t gfp;
 	struct page *new_page;
-	struct mem_cgroup *memcg;
 	pgoff_t index, end = start + HPAGE_PMD_NR;
 	LIST_HEAD(pagelist);
 	XA_STATE_ORDER(xas, &mapping->i_pages, start, HPAGE_PMD_ORDER);
@@ -1534,10 +1632,11 @@ static void collapse_file(struct mm_struct *mm,
 		goto out;
 	}
 
-	if (unlikely(mem_cgroup_try_charge(new_page, mm, gfp, &memcg, true))) {
+	if (unlikely(mem_cgroup_charge(new_page, mm, gfp))) {
 		result = SCAN_CGROUP_CHARGE_FAIL;
 		goto out;
 	}
+	count_memcg_page_event(new_page, THP_COLLAPSE_ALLOC);
 
 	/* This will be less messy when we use multi-index entries */
 	do {
@@ -1547,7 +1646,6 @@ static void collapse_file(struct mm_struct *mm,
 			break;
 		xas_unlock_irq(&xas);
 		if (!xas_nomem(&xas, GFP_KERNEL)) {
-			mem_cgroup_cancel_charge(new_page, memcg, true);
 			result = SCAN_FAIL;
 			goto out;
 		}
@@ -1741,12 +1839,9 @@ out_unlock:
 	}
 
 	if (nr_none) {
-		struct zone *zone = page_zone(new_page);
-
-		__mod_node_page_state(zone->zone_pgdat, NR_FILE_PAGES, nr_none);
+		__mod_lruvec_page_state(new_page, NR_FILE_PAGES, nr_none);
 		if (is_shmem)
-			__mod_node_page_state(zone->zone_pgdat,
-					      NR_SHMEM, nr_none);
+			__mod_lruvec_page_state(new_page, NR_SHMEM, nr_none);
 	}
 
 xa_locked:
@@ -1784,15 +1879,9 @@ xa_unlocked:
 
 		SetPageUptodate(new_page);
 		page_ref_add(new_page, HPAGE_PMD_NR - 1);
-		mem_cgroup_commit_charge(new_page, memcg, false, true);
-
-		if (is_shmem) {
+		if (is_shmem)
 			set_page_dirty(new_page);
-			lru_cache_add_anon(new_page);
-		} else {
-			lru_cache_add_file(new_page);
-		}
-		count_memcg_events(memcg, THP_COLLAPSE_ALLOC, 1);
+		lru_cache_add(new_page);
 
 		/*
 		 * Remove pte page tables, so we can re-fault the page as huge.
@@ -1839,13 +1928,14 @@ xa_unlocked:
 		VM_BUG_ON(nr_none);
 		xas_unlock_irq(&xas);
 
-		mem_cgroup_cancel_charge(new_page, memcg, true);
 		new_page->mapping = NULL;
 	}
 
 	unlock_page(new_page);
 out:
 	VM_BUG_ON(!list_empty(&pagelist));
+	if (!IS_ERR_OR_NULL(*hpage))
+		mem_cgroup_uncharge(*hpage);
 	/* TODO: tracepoints */
 }
 
@@ -2084,6 +2174,8 @@ static void khugepaged_do_scan(void)
 
 	barrier(); /* write khugepaged_pages_to_scan to local stack */
 
+	lru_add_drain_all();
+
 	while (progress < pages) {
 		if (!khugepaged_prealloc_page(&hpage, &wait))
 			break;