Revert "mm: page cache: store only head pages in i_pages"

This reverts commit 5fd4ca2d.

Mikhail Gavrilov reports that it causes the VM_BUG_ON_PAGE() in
__delete_from_swap_cache() to trigger:

   page:ffffd6d34dff0000 refcount:1 mapcount:1 mapping:ffff97812323a689 index:0xfecec363
   anon
   flags: 0x17fffe00080034(uptodate|lru|active|swapbacked)
   raw: 0017fffe00080034 ffffd6d34c67c508 ffffd6d3504b8d48 ffff97812323a689
   raw: 00000000fecec363 0000000000000000 0000000100000000 ffff978433ace000
   page dumped because: VM_BUG_ON_PAGE(entry != page)
   page->mem_cgroup:ffff978433ace000
   ------------[ cut here ]------------
   kernel BUG at mm/swap_state.c:170!
   invalid opcode: 0000 [#1] SMP NOPTI
   CPU: 1 PID: 221 Comm: kswapd0 Not tainted 5.2.0-0.rc2.git0.1.fc31.x86_64 #1
   Hardware name: System manufacturer System Product Name/ROG STRIX X470-I GAMING, BIOS 2202 04/11/2019
   RIP: 0010:__delete_from_swap_cache+0x20d/0x240
   Code: 30 65 48 33 04 25 28 00 00 00 75 4a 48 83 c4 38 5b 5d 41 5c 41 5d 41 5e 41 5f c3 48 c7 c6 2f dc 0f 8a 48 89 c7 e8 93 1b fd ff <0f> 0b 48 c7 c6 a8 74 0f 8a e8 85 1b fd ff 0f 0b 48 c7 c6 a8 7d 0f
   RSP: 0018:ffffa982036e7980 EFLAGS: 00010046
   RAX: 0000000000000021 RBX: 0000000000000040 RCX: 0000000000000006
   RDX: 0000000000000000 RSI: 0000000000000086 RDI: ffff97843d657900
   RBP: 0000000000000001 R08: ffffa982036e7835 R09: 0000000000000535
   R10: ffff97845e21a46c R11: ffffa982036e7835 R12: ffff978426387120
   R13: 0000000000000000 R14: ffffd6d34dff0040 R15: ffffd6d34dff0000
   FS:  0000000000000000(0000) GS:ffff97843d640000(0000) knlGS:0000000000000000
   CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
   CR2: 00002cba88ef5000 CR3: 000000078a97c000 CR4: 00000000003406e0
   Call Trace:
    delete_from_swap_cache+0x46/0xa0
    try_to_free_swap+0xbc/0x110
    swap_writepage+0x13/0x70
    pageout.isra.0+0x13c/0x350
    shrink_page_list+0xc14/0xdf0
    shrink_inactive_list+0x1e5/0x3c0
    shrink_node_memcg+0x202/0x760
    shrink_node+0xe0/0x470
    balance_pgdat+0x2d1/0x510
    kswapd+0x220/0x420
    kthread+0xfb/0x130
    ret_from_fork+0x22/0x40

and it's not immediately obvious why it happens.  It's too late in the
rc cycle to do anything but revert for now.

Link: https://lore.kernel.org/lkml/CABXGCsN9mYmBD-4GaaeW_NrDu+FDXLzr_6x+XNxfmFV6QkYCDg@mail.gmail.com/Reported-and-bisected-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com>
Suggested-by: Jan Kara <jack@suse.cz>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Kirill Shutemov <kirill@shutemov.name>
Cc: William Kucharski <william.kucharski@oracle.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

include/linux/pagemap.h

@@ -333,19 +333,6 @@ static inline struct page *grab_cache_page_nowait(struct address_space *mapping,
 			mapping_gfp_mask(mapping));
 }
 
-static inline struct page *find_subpage(struct page *page, pgoff_t offset)
-{
-	unsigned long mask;
-
-	if (PageHuge(page))
-		return page;
-
-	VM_BUG_ON_PAGE(PageTail(page), page);
-
-	mask = (1UL << compound_order(page)) - 1;
-	return page + (offset & mask);
-}
-
 struct page *find_get_entry(struct address_space *mapping, pgoff_t offset);
 struct page *find_lock_entry(struct address_space *mapping, pgoff_t offset);
 unsigned find_get_entries(struct address_space *mapping, pgoff_t start,

mm/filemap.c

@@ -281,11 +281,11 @@ EXPORT_SYMBOL(delete_from_page_cache);
  * @pvec: pagevec with pages to delete
  *
  * The function walks over mapping->i_pages and removes pages passed in @pvec
- * from the mapping. The function expects @pvec to be sorted by page index
- * and is optimised for it to be dense.
+ * from the mapping. The function expects @pvec to be sorted by page index.
  * It tolerates holes in @pvec (mapping entries at those indices are not
  * modified). The function expects only THP head pages to be present in the
- * @pvec.
+ * @pvec and takes care to delete all corresponding tail pages from the
+ * mapping as well.
  *
  * The function expects the i_pages lock to be held.
  */
@@ -294,44 +294,40 @@ static void page_cache_delete_batch(struct address_space *mapping,
 {
 	XA_STATE(xas, &mapping->i_pages, pvec->pages[0]->index);
 	int total_pages = 0;
-	int i = 0;
+	int i = 0, tail_pages = 0;
 	struct page *page;
 
 	mapping_set_update(&xas, mapping);
 	xas_for_each(&xas, page, ULONG_MAX) {
-		if (i >= pagevec_count(pvec))
+		if (i >= pagevec_count(pvec) && !tail_pages)
 			break;
-
-		/* A swap/dax/shadow entry got inserted? Skip it. */
 		if (xa_is_value(page))
 			continue;
-		/*
-		 * A page got inserted in our range? Skip it. We have our
-		 * pages locked so they are protected from being removed.
-		 * If we see a page whose index is higher than ours, it
-		 * means our page has been removed, which shouldn't be
-		 * possible because we're holding the PageLock.
-		 */
-		if (page != pvec->pages[i]) {
-			VM_BUG_ON_PAGE(page->index > pvec->pages[i]->index,
-					page);
-			continue;
-		}
-
-		WARN_ON_ONCE(!PageLocked(page));
-
-		if (page->index == xas.xa_index)
+		if (!tail_pages) {
+			/*
+			 * Some page got inserted in our range? Skip it. We
+			 * have our pages locked so they are protected from
+			 * being removed.
+			 */
+			if (page != pvec->pages[i]) {
+				VM_BUG_ON_PAGE(page->index >
+						pvec->pages[i]->index, page);
+				continue;
+			}
+			WARN_ON_ONCE(!PageLocked(page));
+			if (PageTransHuge(page) && !PageHuge(page))
+				tail_pages = HPAGE_PMD_NR - 1;
 			page->mapping = NULL;
-		/* Leave page->index set: truncation lookup relies on it */
-
-		/*
-		 * Move to the next page in the vector if this is a regular
-		 * page or the index is of the last sub-page of this compound
-		 * page.
-		 */
-		if (page->index + (1UL << compound_order(page)) - 1 ==
-				xas.xa_index)
+			/*
+			 * Leave page->index set: truncation lookup relies
+			 * upon it
+			 */
 			i++;
+		} else {
+			VM_BUG_ON_PAGE(page->index + HPAGE_PMD_NR - tail_pages
+					!= pvec->pages[i]->index, page);
+			tail_pages--;
+		}
 		xas_store(&xas, NULL);
 		total_pages++;
 	}
@@ -1498,7 +1494,7 @@ EXPORT_SYMBOL(page_cache_prev_miss);
 struct page *find_get_entry(struct address_space *mapping, pgoff_t offset)
 {
 	XA_STATE(xas, &mapping->i_pages, offset);
-	struct page *page;
+	struct page *head, *page;
 
 	rcu_read_lock();
 repeat:
@@ -1513,19 +1509,25 @@ struct page *find_get_entry(struct address_space *mapping, pgoff_t offset)
 	if (!page || xa_is_value(page))
 		goto out;
 
-	if (!page_cache_get_speculative(page))
+	head = compound_head(page);
+	if (!page_cache_get_speculative(head))
+		goto repeat;
+
+	/* The page was split under us? */
+	if (compound_head(page) != head) {
+		put_page(head);
 		goto repeat;
+	}
 
 	/*
-	 * Has the page moved or been split?
+	 * Has the page moved?
 	 * This is part of the lockless pagecache protocol. See
 	 * include/linux/pagemap.h for details.
 	 */
 	if (unlikely(page != xas_reload(&xas))) {
-		put_page(page);
+		put_page(head);
 		goto repeat;
 	}
-	page = find_subpage(page, offset);
 out:
 	rcu_read_unlock();
 
@@ -1707,6 +1709,7 @@ unsigned find_get_entries(struct address_space *mapping,
 
 	rcu_read_lock();
 	xas_for_each(&xas, page, ULONG_MAX) {
+		struct page *head;
 		if (xas_retry(&xas, page))
 			continue;
 		/*
@@ -1717,13 +1720,17 @@ unsigned find_get_entries(struct address_space *mapping,
 		if (xa_is_value(page))
 			goto export;
 
-		if (!page_cache_get_speculative(page))
+		head = compound_head(page);
+		if (!page_cache_get_speculative(head))
 			goto retry;
 
-		/* Has the page moved or been split? */
+		/* The page was split under us? */
+		if (compound_head(page) != head)
+			goto put_page;
+
+		/* Has the page moved? */
 		if (unlikely(page != xas_reload(&xas)))
 			goto put_page;
-		page = find_subpage(page, xas.xa_index);
 
 export:
 		indices[ret] = xas.xa_index;
@@ -1732,7 +1739,7 @@ unsigned find_get_entries(struct address_space *mapping,
 			break;
 		continue;
 put_page:
-		put_page(page);
+		put_page(head);
 retry:
 		xas_reset(&xas);
 	}
@@ -1774,27 +1781,33 @@ unsigned find_get_pages_range(struct address_space *mapping, pgoff_t *start,
 
 	rcu_read_lock();
 	xas_for_each(&xas, page, end) {
+		struct page *head;
 		if (xas_retry(&xas, page))
 			continue;
 		/* Skip over shadow, swap and DAX entries */
 		if (xa_is_value(page))
 			continue;
 
-		if (!page_cache_get_speculative(page))
+		head = compound_head(page);
+		if (!page_cache_get_speculative(head))
 			goto retry;
 
-		/* Has the page moved or been split? */
+		/* The page was split under us? */
+		if (compound_head(page) != head)
+			goto put_page;
+
+		/* Has the page moved? */
 		if (unlikely(page != xas_reload(&xas)))
 			goto put_page;
 
-		pages[ret] = find_subpage(page, xas.xa_index);
+		pages[ret] = page;
 		if (++ret == nr_pages) {
 			*start = xas.xa_index + 1;
 			goto out;
 		}
 		continue;
 put_page:
-		put_page(page);
+		put_page(head);
 retry:
 		xas_reset(&xas);
 	}
@@ -1839,6 +1852,7 @@ unsigned find_get_pages_contig(struct address_space *mapping, pgoff_t index,
 
 	rcu_read_lock();
 	for (page = xas_load(&xas); page; page = xas_next(&xas)) {
+		struct page *head;
 		if (xas_retry(&xas, page))
 			continue;
 		/*
@@ -1848,19 +1862,24 @@ unsigned find_get_pages_contig(struct address_space *mapping, pgoff_t index,
 		if (xa_is_value(page))
 			break;
 
-		if (!page_cache_get_speculative(page))
+		head = compound_head(page);
+		if (!page_cache_get_speculative(head))
 			goto retry;
 
-		/* Has the page moved or been split? */
+		/* The page was split under us? */
+		if (compound_head(page) != head)
+			goto put_page;
+
+		/* Has the page moved? */
 		if (unlikely(page != xas_reload(&xas)))
 			goto put_page;
 
-		pages[ret] = find_subpage(page, xas.xa_index);
+		pages[ret] = page;
 		if (++ret == nr_pages)
 			break;
 		continue;
 put_page:
-		put_page(page);
+		put_page(head);
 retry:
 		xas_reset(&xas);
 	}
@@ -1896,6 +1915,7 @@ unsigned find_get_pages_range_tag(struct address_space *mapping, pgoff_t *index,
 
 	rcu_read_lock();
 	xas_for_each_marked(&xas, page, end, tag) {
+		struct page *head;
 		if (xas_retry(&xas, page))
 			continue;
 		/*
@@ -1906,21 +1926,26 @@ unsigned find_get_pages_range_tag(struct address_space *mapping, pgoff_t *index,
 		if (xa_is_value(page))
 			continue;
 
-		if (!page_cache_get_speculative(page))
+		head = compound_head(page);
+		if (!page_cache_get_speculative(head))
 			goto retry;
 
-		/* Has the page moved or been split? */
+		/* The page was split under us? */
+		if (compound_head(page) != head)
+			goto put_page;
+
+		/* Has the page moved? */
 		if (unlikely(page != xas_reload(&xas)))
 			goto put_page;
 
-		pages[ret] = find_subpage(page, xas.xa_index);
+		pages[ret] = page;
 		if (++ret == nr_pages) {
 			*index = xas.xa_index + 1;
 			goto out;
 		}
 		continue;
 put_page:
-		put_page(page);
+		put_page(head);
 retry:
 		xas_reset(&xas);
 	}
@@ -2603,7 +2628,7 @@ void filemap_map_pages(struct vm_fault *vmf,
 	pgoff_t last_pgoff = start_pgoff;
 	unsigned long max_idx;
 	XA_STATE(xas, &mapping->i_pages, start_pgoff);
-	struct page *page;
+	struct page *head, *page;
 
 	rcu_read_lock();
 	xas_for_each(&xas, page, end_pgoff) {
@@ -2612,19 +2637,24 @@ void filemap_map_pages(struct vm_fault *vmf,
 		if (xa_is_value(page))
 			goto next;
 
+		head = compound_head(page);
+
 		/*
 		 * Check for a locked page first, as a speculative
 		 * reference may adversely influence page migration.
 		 */
-		if (PageLocked(page))
+		if (PageLocked(head))
 			goto next;
-		if (!page_cache_get_speculative(page))
+		if (!page_cache_get_speculative(head))
 			goto next;
 
-		/* Has the page moved or been split? */
+		/* The page was split under us? */
+		if (compound_head(page) != head)
+			goto skip;
+
+		/* Has the page moved? */
 		if (unlikely(page != xas_reload(&xas)))
 			goto skip;
-		page = find_subpage(page, xas.xa_index);
 
 		if (!PageUptodate(page) ||
 				PageReadahead(page) ||

mm/huge_memory.c

@@ -2496,9 +2496,6 @@ static void __split_huge_page(struct page *page, struct list_head *list,
 			if (IS_ENABLED(CONFIG_SHMEM) && PageSwapBacked(head))
 				shmem_uncharge(head->mapping->host, 1);
 			put_page(head + i);
-		} else if (!PageAnon(page)) {
-			__xa_store(&head->mapping->i_pages, head[i].index,
-					head + i, 0);
 		}
 	}
 

mm/khugepaged.c

@@ -1378,7 +1378,7 @@ static void collapse_shmem(struct mm_struct *mm,
 				result = SCAN_FAIL;
 				goto xa_locked;
 			}
-			xas_store(&xas, new_page);
+			xas_store(&xas, new_page + (index % HPAGE_PMD_NR));
 			nr_none++;
 			continue;
 		}
@@ -1454,7 +1454,7 @@ static void collapse_shmem(struct mm_struct *mm,
 		list_add_tail(&page->lru, &pagelist);
 
 		/* Finally, replace with the new page. */
-		xas_store(&xas, new_page);
+		xas_store(&xas, new_page + (index % HPAGE_PMD_NR));
 		continue;
 out_unlock:
 		unlock_page(page);

mm/memfd.c

@@ -39,7 +39,6 @@ static void memfd_tag_pins(struct xa_state *xas)
 	xas_for_each(xas, page, ULONG_MAX) {
 		if (xa_is_value(page))
 			continue;
-		page = find_subpage(page, xas->xa_index);
 		if (page_count(page) - page_mapcount(page) > 1)
 			xas_set_mark(xas, MEMFD_TAG_PINNED);
 
@@ -89,7 +88,6 @@ static int memfd_wait_for_pins(struct address_space *mapping)
 			bool clear = true;
 			if (xa_is_value(page))
 				continue;
-			page = find_subpage(page, xas.xa_index);
 			if (page_count(page) - page_mapcount(page) != 1) {
 				/*
 				 * On the last scan, we clean up all those tags

mm/migrate.c

@@ -463,7 +463,7 @@ int migrate_page_move_mapping(struct address_space *mapping,
 
 		for (i = 1; i < HPAGE_PMD_NR; i++) {
 			xas_next(&xas);
-			xas_store(&xas, newpage);
+			xas_store(&xas, newpage + i);
 		}
 	}
 

mm/shmem.c

@@ -614,7 +614,7 @@ static int shmem_add_to_page_cache(struct page *page,
 		if (xas_error(&xas))
 			goto unlock;
 next:
-		xas_store(&xas, page);
+		xas_store(&xas, page + i);
 		if (++i < nr) {
 			xas_next(&xas);
 			goto next;

mm/swap_state.c

@@ -132,7 +132,7 @@ int add_to_swap_cache(struct page *page, swp_entry_t entry, gfp_t gfp)
 		for (i = 0; i < nr; i++) {
 			VM_BUG_ON_PAGE(xas.xa_index != idx + i, page);
 			set_page_private(page + i, entry.val + i);
-			xas_store(&xas, page);
+			xas_store(&xas, page + i);
 			xas_next(&xas);
 		}
 		address_space->nrpages += nr;
@@ -167,7 +167,7 @@ void __delete_from_swap_cache(struct page *page, swp_entry_t entry)
 
 	for (i = 0; i < nr; i++) {
 		void *entry = xas_store(&xas, NULL);
-		VM_BUG_ON_PAGE(entry != page, entry);
+		VM_BUG_ON_PAGE(entry != page + i, entry);
 		set_page_private(page + i, 0);
 		xas_next(&xas);
 	}