Merge tag 'ext4_for_linus-6.8-rc1' of git://git.kernel.org/pub/scm/li…

…nux/kernel/git/tytso/ext4

Pull ext4 updates from Ted Ts'o:
 "Various ext4 bug fixes and cleanups. The fixes are mostly in the
  fstrim and mballoc code paths.

  Also enable dioread_nolock in the case where the block size is less
  than the page size (dioread_nolock has been default in the bs == ps
  case for quite some time)"

* tag 'ext4_for_linus-6.8-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
  ext4: fix inconsistent between segment fstrim and full fstrim
  ext4: fallback to complex scan if aligned scan doesn't work
  ext4: convert ext4_da_do_write_end() to take a folio
  ext4: allow for the last group to be marked as trimmed
  ext4: move ext4_check_bdev_write_error() into nojournal mode
  jbd2: abort journal when detecting metadata writeback error of fs dev
  jbd2: remove unused 'JBD2_CHECKPOINT_IO_ERROR' and 'j_atomic_flags'
  jbd2: replace journal state flag by checking errseq
  jbd2: add errseq to detect client fs's bdev writeback error
  ext4: improving calculation of 'fe_{len|start}' in mb_find_extent()
  ext4: clarify handling of unwritten bh in __ext4_block_zero_page_range()
  ext4: treat end of range as exclusive in ext4_zero_range()
  ext4: enable dioread_nolock as default for bs < ps case
  ext4: delete redundant calculations in ext4_mb_get_buddy_page_lock()
  ext4: reduce unnecessary memory allocation in alloc_flex_gd()
  ext4: avoid online resizing failures due to oversized flex bg
  ext4: remove unnecessary check from alloc_flex_gd()
  ext4: unify the type of flexbg_size to unsigned int

fs/ext4/ext4_jbd2.c

@@ -235,16 +235,15 @@ int __ext4_journal_get_write_access(const char *where, unsigned int line,
 
 	might_sleep();
 
-	ext4_check_bdev_write_error(sb);
-
 	if (ext4_handle_valid(handle)) {
 		err = jbd2_journal_get_write_access(handle, bh);
 		if (err) {
 			ext4_journal_abort_handle(where, line, __func__, bh,
 						  handle, err);
 			return err;
 		}
-	}
+	} else
+		ext4_check_bdev_write_error(sb);
 	if (trigger_type == EXT4_JTR_NONE || !ext4_has_metadata_csum(sb))
 		return 0;
 	BUG_ON(trigger_type >= EXT4_JOURNAL_TRIGGER_COUNT);

fs/ext4/extents.c

@@ -4523,7 +4523,8 @@ static long ext4_zero_range(struct file *file, loff_t offset,
 	 * Round up offset. This is not fallocate, we need to zero out
 	 * blocks, so convert interior block aligned part of the range to
 	 * unwritten and possibly manually zero out unaligned parts of the
-	 * range.
+	 * range. Here, start and partial_begin are inclusive, end and
+	 * partial_end are exclusive.
 	 */
 	start = round_up(offset, 1 << blkbits);
 	end = round_down((offset + len), 1 << blkbits);
@@ -4609,7 +4610,8 @@ static long ext4_zero_range(struct file *file, loff_t offset,
 		 * disk in case of crash before zeroing trans is committed.
 		 */
 		if (ext4_should_journal_data(inode)) {
-			ret = filemap_write_and_wait_range(mapping, start, end);
+			ret = filemap_write_and_wait_range(mapping, start,
+							   end - 1);
 			if (ret) {
 				filemap_invalidate_unlock(mapping);
 				goto out_mutex;

fs/ext4/inode.c

@@ -2947,7 +2947,7 @@ static int ext4_da_should_update_i_disksize(struct folio *folio,
 
 static int ext4_da_do_write_end(struct address_space *mapping,
 			loff_t pos, unsigned len, unsigned copied,
-			struct page *page)
+			struct folio *folio)
 {
 	struct inode *inode = mapping->host;
 	loff_t old_size = inode->i_size;
@@ -2958,12 +2958,13 @@ static int ext4_da_do_write_end(struct address_space *mapping,
 	 * block_write_end() will mark the inode as dirty with I_DIRTY_PAGES
 	 * flag, which all that's needed to trigger page writeback.
 	 */
-	copied = block_write_end(NULL, mapping, pos, len, copied, page, NULL);
+	copied = block_write_end(NULL, mapping, pos, len, copied,
+			&folio->page, NULL);
 	new_i_size = pos + copied;
 
 	/*
-	 * It's important to update i_size while still holding page lock,
-	 * because page writeout could otherwise come in and zero beyond
+	 * It's important to update i_size while still holding folio lock,
+	 * because folio writeout could otherwise come in and zero beyond
 	 * i_size.
 	 *
 	 * Since we are holding inode lock, we are sure i_disksize <=
@@ -2981,14 +2982,14 @@ static int ext4_da_do_write_end(struct address_space *mapping,
 
 		i_size_write(inode, new_i_size);
 		end = (new_i_size - 1) & (PAGE_SIZE - 1);
-		if (copied && ext4_da_should_update_i_disksize(page_folio(page), end)) {
+		if (copied && ext4_da_should_update_i_disksize(folio, end)) {
 			ext4_update_i_disksize(inode, new_i_size);
 			disksize_changed = true;
 		}
 	}
 
-	unlock_page(page);
-	put_page(page);
+	folio_unlock(folio);
+	folio_put(folio);
 
 	if (old_size < pos)
 		pagecache_isize_extended(inode, old_size, pos);
@@ -3027,10 +3028,10 @@ static int ext4_da_write_end(struct file *file,
 		return ext4_write_inline_data_end(inode, pos, len, copied,
 						  folio);
 
-	if (unlikely(copied < len) && !PageUptodate(page))
+	if (unlikely(copied < len) && !folio_test_uptodate(folio))
 		copied = 0;
 
-	return ext4_da_do_write_end(mapping, pos, len, copied, &folio->page);
+	return ext4_da_do_write_end(mapping, pos, len, copied, folio);
 }
 
 /*
@@ -3630,6 +3631,12 @@ void ext4_set_aops(struct inode *inode)
 		inode->i_mapping->a_ops = &ext4_aops;
 }
 
+/*
+ * Here we can't skip an unwritten buffer even though it usually reads zero
+ * because it might have data in pagecache (eg, if called from ext4_zero_range,
+ * ext4_punch_hole, etc) which needs to be properly zeroed out. Otherwise a
+ * racing writeback can come later and flush the stale pagecache to disk.
+ */
 static int __ext4_block_zero_page_range(handle_t *handle,
 		struct address_space *mapping, loff_t from, loff_t length)
 {

fs/ext4/mballoc.c

@@ -1456,9 +1456,8 @@ static int ext4_mb_get_buddy_page_lock(struct super_block *sb,
 		return 0;
 	}
 
-	block++;
-	pnum = block / blocks_per_page;
-	page = find_or_create_page(inode->i_mapping, pnum, gfp);
+	/* blocks_per_page == 1, hence we need another page for the buddy */
+	page = find_or_create_page(inode->i_mapping, block + 1, gfp);
 	if (!page)
 		return -ENOMEM;
 	BUG_ON(page->mapping != inode->i_mapping);
@@ -1958,8 +1957,7 @@ static void mb_free_blocks(struct inode *inode, struct ext4_buddy *e4b,
 static int mb_find_extent(struct ext4_buddy *e4b, int block,
 				int needed, struct ext4_free_extent *ex)
 {
-	int next = block;
-	int max, order;
+	int max, order, next;
 	void *buddy;
 
 	assert_spin_locked(ext4_group_lock_ptr(e4b->bd_sb, e4b->bd_group));
@@ -1977,16 +1975,12 @@ static int mb_find_extent(struct ext4_buddy *e4b, int block,
 
 	/* find actual order */
 	order = mb_find_order_for_block(e4b, block);
-	block = block >> order;
 
-	ex->fe_len = 1 << order;
-	ex->fe_start = block << order;
+	ex->fe_len = (1 << order) - (block & ((1 << order) - 1));
+	ex->fe_start = block;
 	ex->fe_group = e4b->bd_group;
 
-	/* calc difference from given start */
-	next = next - ex->fe_start;
-	ex->fe_len -= next;
-	ex->fe_start += next;
+	block = block >> order;
 
 	while (needed > ex->fe_len &&
 	       mb_find_buddy(e4b, order, &max)) {
@@ -2895,14 +2889,19 @@ ext4_mb_regular_allocator(struct ext4_allocation_context *ac)
 			ac->ac_groups_scanned++;
 			if (cr == CR_POWER2_ALIGNED)
 				ext4_mb_simple_scan_group(ac, &e4b);
-			else if ((cr == CR_GOAL_LEN_FAST ||
-				 cr == CR_BEST_AVAIL_LEN) &&
-				 sbi->s_stripe &&
-				 !(ac->ac_g_ex.fe_len %
-				 EXT4_B2C(sbi, sbi->s_stripe)))
-				ext4_mb_scan_aligned(ac, &e4b);
-			else
-				ext4_mb_complex_scan_group(ac, &e4b);
+			else {
+				bool is_stripe_aligned = sbi->s_stripe &&
+					!(ac->ac_g_ex.fe_len %
+					  EXT4_B2C(sbi, sbi->s_stripe));
+
+				if ((cr == CR_GOAL_LEN_FAST ||
+				     cr == CR_BEST_AVAIL_LEN) &&
+				    is_stripe_aligned)
+					ext4_mb_scan_aligned(ac, &e4b);
+
+				if (ac->ac_status == AC_STATUS_CONTINUE)
+					ext4_mb_complex_scan_group(ac, &e4b);
+			}
 
 			ext4_unlock_group(sb, group);
 			ext4_mb_unload_buddy(&e4b);
@@ -6735,11 +6734,16 @@ __acquires(bitlock)
 static ext4_grpblk_t ext4_last_grp_cluster(struct super_block *sb,
 					   ext4_group_t grp)
 {
-	if (grp < ext4_get_groups_count(sb))
-		return EXT4_CLUSTERS_PER_GROUP(sb) - 1;
-	return (ext4_blocks_count(EXT4_SB(sb)->s_es) -
-		ext4_group_first_block_no(sb, grp) - 1) >>
-					EXT4_CLUSTER_BITS(sb);
+	unsigned long nr_clusters_in_group;
+
+	if (grp < (ext4_get_groups_count(sb) - 1))
+		nr_clusters_in_group = EXT4_CLUSTERS_PER_GROUP(sb);
+	else
+		nr_clusters_in_group = (ext4_blocks_count(EXT4_SB(sb)->s_es) -
+					ext4_group_first_block_no(sb, grp))
+				       >> EXT4_CLUSTER_BITS(sb);
+
+	return nr_clusters_in_group - 1;
 }
 
 static bool ext4_trim_interrupted(void)
@@ -6753,13 +6757,15 @@ static int ext4_try_to_trim_range(struct super_block *sb,
 __acquires(ext4_group_lock_ptr(sb, e4b->bd_group))
 __releases(ext4_group_lock_ptr(sb, e4b->bd_group))
 {
-	ext4_grpblk_t next, count, free_count;
+	ext4_grpblk_t next, count, free_count, last, origin_start;
 	bool set_trimmed = false;
 	void *bitmap;
 
+	last = ext4_last_grp_cluster(sb, e4b->bd_group);
 	bitmap = e4b->bd_bitmap;
-	if (start == 0 && max >= ext4_last_grp_cluster(sb, e4b->bd_group))
+	if (start == 0 && max >= last)
 		set_trimmed = true;
+	origin_start = start;
 	start = max(e4b->bd_info->bb_first_free, start);
 	count = 0;
 	free_count = 0;
@@ -6768,7 +6774,10 @@ __releases(ext4_group_lock_ptr(sb, e4b->bd_group))
 		start = mb_find_next_zero_bit(bitmap, max + 1, start);
 		if (start > max)
 			break;
-		next = mb_find_next_bit(bitmap, max + 1, start);
+
+		next = mb_find_next_bit(bitmap, last + 1, start);
+		if (origin_start == 0 && next >= last)
+			set_trimmed = true;
 
 		if ((next - start) >= minblocks) {
 			int ret = ext4_trim_extent(sb, start, next - start, e4b);

fs/ext4/resize.c

@@ -218,35 +218,53 @@ struct ext4_new_flex_group_data {
 						   in the flex group */
 	__u16 *bg_flags;			/* block group flags of groups
 						   in @groups */
+	ext4_group_t resize_bg;			/* number of allocated
+						   new_group_data */
 	ext4_group_t count;			/* number of groups in @groups
 						 */
 };
 
+/*
+ * Avoiding memory allocation failures due to too many groups added each time.
+ */
+#define MAX_RESIZE_BG				16384
+
 /*
  * alloc_flex_gd() allocates a ext4_new_flex_group_data with size of
  * @flexbg_size.
  *
  * Returns NULL on failure otherwise address of the allocated structure.
  */
-static struct ext4_new_flex_group_data *alloc_flex_gd(unsigned long flexbg_size)
+static struct ext4_new_flex_group_data *alloc_flex_gd(unsigned int flexbg_size,
+				ext4_group_t o_group, ext4_group_t n_group)
 {
+	ext4_group_t last_group;
 	struct ext4_new_flex_group_data *flex_gd;
 
 	flex_gd = kmalloc(sizeof(*flex_gd), GFP_NOFS);
 	if (flex_gd == NULL)
 		goto out3;
 
-	if (flexbg_size >= UINT_MAX / sizeof(struct ext4_new_group_data))
-		goto out2;
-	flex_gd->count = flexbg_size;
+	if (unlikely(flexbg_size > MAX_RESIZE_BG))
+		flex_gd->resize_bg = MAX_RESIZE_BG;
+	else
+		flex_gd->resize_bg = flexbg_size;
+
+	/* Avoid allocating large 'groups' array if not needed */
+	last_group = o_group | (flex_gd->resize_bg - 1);
+	if (n_group <= last_group)
+		flex_gd->resize_bg = 1 << fls(n_group - o_group + 1);
+	else if (n_group - last_group < flex_gd->resize_bg)
+		flex_gd->resize_bg = 1 << max(fls(last_group - o_group + 1),
+					      fls(n_group - last_group));
 
-	flex_gd->groups = kmalloc_array(flexbg_size,
+	flex_gd->groups = kmalloc_array(flex_gd->resize_bg,
 					sizeof(struct ext4_new_group_data),
 					GFP_NOFS);
 	if (flex_gd->groups == NULL)
 		goto out2;
 
-	flex_gd->bg_flags = kmalloc_array(flexbg_size, sizeof(__u16),
+	flex_gd->bg_flags = kmalloc_array(flex_gd->resize_bg, sizeof(__u16),
 					  GFP_NOFS);
 	if (flex_gd->bg_flags == NULL)
 		goto out1;
@@ -283,7 +301,7 @@ static void free_flex_gd(struct ext4_new_flex_group_data *flex_gd)
  */
 static int ext4_alloc_group_tables(struct super_block *sb,
 				struct ext4_new_flex_group_data *flex_gd,
-				int flexbg_size)
+				unsigned int flexbg_size)
 {
 	struct ext4_new_group_data *group_data = flex_gd->groups;
 	ext4_fsblk_t start_blk;
@@ -384,12 +402,12 @@ static int ext4_alloc_group_tables(struct super_block *sb,
 		group = group_data[0].group;
 
 		printk(KERN_DEBUG "EXT4-fs: adding a flex group with "
-		       "%d groups, flexbg size is %d:\n", flex_gd->count,
+		       "%u groups, flexbg size is %u:\n", flex_gd->count,
 		       flexbg_size);
 
 		for (i = 0; i < flex_gd->count; i++) {
 			ext4_debug(
-			       "adding %s group %u: %u blocks (%d free, %d mdata blocks)\n",
+			       "adding %s group %u: %u blocks (%u free, %u mdata blocks)\n",
 			       ext4_bg_has_super(sb, group + i) ? "normal" :
 			       "no-super", group + i,
 			       group_data[i].blocks_count,
@@ -1605,8 +1623,7 @@ static int ext4_flex_group_add(struct super_block *sb,
 
 static int ext4_setup_next_flex_gd(struct super_block *sb,
 				    struct ext4_new_flex_group_data *flex_gd,
-				    ext4_fsblk_t n_blocks_count,
-				    unsigned long flexbg_size)
+				    ext4_fsblk_t n_blocks_count)
 {
 	struct ext4_sb_info *sbi = EXT4_SB(sb);
 	struct ext4_super_block *es = sbi->s_es;
@@ -1630,7 +1647,7 @@ static int ext4_setup_next_flex_gd(struct super_block *sb,
 	BUG_ON(last);
 	ext4_get_group_no_and_offset(sb, n_blocks_count - 1, &n_group, &last);
 
-	last_group = group | (flexbg_size - 1);
+	last_group = group | (flex_gd->resize_bg - 1);
 	if (last_group > n_group)
 		last_group = n_group;
 
@@ -1990,8 +2007,9 @@ int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count)
 	ext4_fsblk_t o_blocks_count;
 	ext4_fsblk_t n_blocks_count_retry = 0;
 	unsigned long last_update_time = 0;
-	int err = 0, flexbg_size = 1 << sbi->s_log_groups_per_flex;
+	int err = 0;
 	int meta_bg;
+	unsigned int flexbg_size = ext4_flex_bg_size(sbi);
 
 	/* See if the device is actually as big as what was requested */
 	bh = ext4_sb_bread(sb, n_blocks_count - 1, 0);
@@ -2123,7 +2141,7 @@ int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count)
 	if (err)
 		goto out;
 
-	flex_gd = alloc_flex_gd(flexbg_size);
+	flex_gd = alloc_flex_gd(flexbg_size, o_group, n_group);
 	if (flex_gd == NULL) {
 		err = -ENOMEM;
 		goto out;
@@ -2132,8 +2150,7 @@ int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count)
 	/* Add flex groups. Note that a regular group is a
 	 * flex group with 1 group.
 	 */
-	while (ext4_setup_next_flex_gd(sb, flex_gd, n_blocks_count,
-					      flexbg_size)) {
+	while (ext4_setup_next_flex_gd(sb, flex_gd, n_blocks_count)) {
 		if (time_is_before_jiffies(last_update_time + HZ * 10)) {
 			if (last_update_time)
 				ext4_msg(sb, KERN_INFO,

fs/ext4/super.c

@@ -2793,15 +2793,6 @@ static int ext4_check_opt_consistency(struct fs_context *fc,
 		return -EINVAL;
 	}
 
-	if (ctx_test_mount_opt(ctx, EXT4_MOUNT_DIOREAD_NOLOCK)) {
-		int blocksize =
-			BLOCK_SIZE << le32_to_cpu(sbi->s_es->s_log_block_size);
-		if (blocksize < PAGE_SIZE)
-			ext4_msg(NULL, KERN_WARNING, "Warning: mounting with an "
-				 "experimental mount option 'dioread_nolock' "
-				 "for blocksize < PAGE_SIZE");
-	}
-
 	err = ext4_check_test_dummy_encryption(fc, sb);
 	if (err)
 		return err;
@@ -4410,7 +4401,7 @@ static void ext4_set_def_opts(struct super_block *sb,
 	    ((def_mount_opts & EXT4_DEFM_NODELALLOC) == 0))
 		set_opt(sb, DELALLOC);
 
-	if (sb->s_blocksize == PAGE_SIZE)
+	if (sb->s_blocksize <= PAGE_SIZE)
 		set_opt(sb, DIOREAD_NOLOCK);
 }