[ 36/73] ext4: fix overhead calculation used by ext4_statfs()

[Ben Hutchings]

3.2-stable review patch.  If anyone has any objections, please let me know.

------------------

From: Theodore Ts'o <tytso@xxxxxxx>

commit 952fc18ef9ec707ebdc16c0786ec360295e5ff15 upstream.

Commit f975d6bcc7a introduced bug which caused ext4_statfs() to
miscalculate the number of file system overhead blocks.  This causes
the f_blocks field in the statfs structure to be larger than it should
be.  This would in turn cause the "df" output to show the number of
data blocks in the file system and the number of data blocks used to
be larger than they should be.

Signed-off-by: "Theodore Ts'o" <tytso@xxxxxxx>
[bwh: Backported to 3.2: adjust context]
Signed-off-by: Ben Hutchings <ben@xxxxxxxxxxxxxxx>
---
I put the call to ext4_calculate_overhead() in ext4_group_add(); is
it the right place?

Ben.
---
 fs/ext4/bitmap.c |    4 --
 fs/ext4/ext4.h   |    4 +-
 fs/ext4/resize.c |    7 ++-
 fs/ext4/super.c  |  174 ++++++++++++++++++++++++++++++++++++++----------------
 4 files changed, 132 insertions(+), 57 deletions(-)

--- a/fs/ext4/bitmap.c
+++ b/fs/ext4/bitmap.c
@@ -11,8 +11,6 @@
 #include <linux/jbd2.h>
 #include "ext4.h"
 
-#ifdef EXT4FS_DEBUG
-
 static const int nibblemap[] = {4, 3, 3, 2, 3, 2, 2, 1, 3, 2, 2, 1, 2, 1, 1, 0};
 
 unsigned int ext4_count_free(char *bitmap, unsigned int numchars)
@@ -25,5 +23,3 @@ unsigned int ext4_count_free(char *bitma
 	return sum;
 }
 
-#endif  /*  EXT4FS_DEBUG  */
-
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -1123,8 +1123,7 @@ struct ext4_sb_info {
 	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
 	ext4_group_t s_groups_count;	/* Number of groups in the fs */
 	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
-	unsigned long s_overhead_last;  /* Last calculated overhead */
-	unsigned long s_blocks_last;    /* Last seen block count */
+	unsigned long s_overhead;  /* # of fs overhead clusters */
 	unsigned int s_cluster_ratio;	/* Number of blocks per cluster */
 	unsigned int s_cluster_bits;	/* log2 of s_cluster_ratio */
 	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
@@ -1925,6 +1924,7 @@ extern int ext4_group_extend(struct supe
 				ext4_fsblk_t n_blocks_count);
 
 /* super.c */
+extern int ext4_calculate_overhead(struct super_block *sb);
 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
 extern void ext4_kvfree(void *ptr);
--- a/fs/ext4/resize.c
+++ b/fs/ext4/resize.c
@@ -952,6 +952,11 @@ int ext4_group_add(struct super_block *s
 			   &sbi->s_flex_groups[flex_group].free_inodes);
 	}
 
+	/*
+	 * Update the fs overhead information
+	 */
+	ext4_calculate_overhead(sb);
+
 	ext4_handle_dirty_super(handle, sb);
 
 exit_journal:
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -3083,6 +3083,114 @@ static void ext4_destroy_lazyinit_thread
 	kthread_stop(ext4_lazyinit_task);
 }
 
+/*
+ * Note: calculating the overhead so we can be compatible with
+ * historical BSD practice is quite difficult in the face of
+ * clusters/bigalloc.  This is because multiple metadata blocks from
+ * different block group can end up in the same allocation cluster.
+ * Calculating the exact overhead in the face of clustered allocation
+ * requires either O(all block bitmaps) in memory or O(number of block
+ * groups**2) in time.  We will still calculate the superblock for
+ * older file systems --- and if we come across with a bigalloc file
+ * system with zero in s_overhead_clusters the estimate will be close to
+ * correct especially for very large cluster sizes --- but for newer
+ * file systems, it's better to calculate this figure once at mkfs
+ * time, and store it in the superblock.  If the superblock value is
+ * present (even for non-bigalloc file systems), we will use it.
+ */
+static int count_overhead(struct super_block *sb, ext4_group_t grp,
+			  char *buf)
+{
+	struct ext4_sb_info	*sbi = EXT4_SB(sb);
+	struct ext4_group_desc	*gdp;
+	ext4_fsblk_t		first_block, last_block, b;
+	ext4_group_t		i, ngroups = ext4_get_groups_count(sb);
+	int			s, j, count = 0;
+
+	first_block = le32_to_cpu(sbi->s_es->s_first_data_block) +
+		(grp * EXT4_BLOCKS_PER_GROUP(sb));
+	last_block = first_block + EXT4_BLOCKS_PER_GROUP(sb) - 1;
+	for (i = 0; i < ngroups; i++) {
+		gdp = ext4_get_group_desc(sb, i, NULL);
+		b = ext4_block_bitmap(sb, gdp);
+		if (b >= first_block && b <= last_block) {
+			ext4_set_bit(EXT4_B2C(sbi, b - first_block), buf);
+			count++;
+		}
+		b = ext4_inode_bitmap(sb, gdp);
+		if (b >= first_block && b <= last_block) {
+			ext4_set_bit(EXT4_B2C(sbi, b - first_block), buf);
+			count++;
+		}
+		b = ext4_inode_table(sb, gdp);
+		if (b >= first_block && b + sbi->s_itb_per_group <= last_block)
+			for (j = 0; j < sbi->s_itb_per_group; j++, b++) {
+				int c = EXT4_B2C(sbi, b - first_block);
+				ext4_set_bit(c, buf);
+				count++;
+			}
+		if (i != grp)
+			continue;
+		s = 0;
+		if (ext4_bg_has_super(sb, grp)) {
+			ext4_set_bit(s++, buf);
+			count++;
+		}
+		for (j = ext4_bg_num_gdb(sb, grp); j > 0; j--) {
+			ext4_set_bit(EXT4_B2C(sbi, s++), buf);
+			count++;
+		}
+	}
+	if (!count)
+		return 0;
+	return EXT4_CLUSTERS_PER_GROUP(sb) -
+		ext4_count_free(buf, EXT4_CLUSTERS_PER_GROUP(sb) / 8);
+}
+
+/*
+ * Compute the overhead and stash it in sbi->s_overhead
+ */
+int ext4_calculate_overhead(struct super_block *sb)
+{
+	struct ext4_sb_info *sbi = EXT4_SB(sb);
+	struct ext4_super_block *es = sbi->s_es;
+	ext4_group_t i, ngroups = ext4_get_groups_count(sb);
+	ext4_fsblk_t overhead = 0;
+	char *buf = (char *) get_zeroed_page(GFP_KERNEL);
+
+	memset(buf, 0, PAGE_SIZE);
+	if (!buf)
+		return -ENOMEM;
+
+	/*
+	 * Compute the overhead (FS structures).  This is constant
+	 * for a given filesystem unless the number of block groups
+	 * changes so we cache the previous value until it does.
+	 */
+
+	/*
+	 * All of the blocks before first_data_block are overhead
+	 */
+	overhead = EXT4_B2C(sbi, le32_to_cpu(es->s_first_data_block));
+
+	/*
+	 * Add the overhead found in each block group
+	 */
+	for (i = 0; i < ngroups; i++) {
+		int blks;
+
+		blks = count_overhead(sb, i, buf);
+		overhead += blks;
+		if (blks)
+			memset(buf, 0, PAGE_SIZE);
+		cond_resched();
+	}
+	sbi->s_overhead = overhead;
+	smp_wmb();
+	free_page((unsigned long) buf);
+	return 0;
+}
+
 static int ext4_fill_super(struct super_block *sb, void *data, int silent)
 {
 	char *orig_data = kstrdup(data, GFP_KERNEL);
@@ -3695,6 +3803,18 @@ static int ext4_fill_super(struct super_
 
 no_journal:
 	/*
+	 * Get the # of file system overhead blocks from the
+	 * superblock if present.
+	 */
+	if (es->s_overhead_clusters)
+		sbi->s_overhead = le32_to_cpu(es->s_overhead_clusters);
+	else {
+		ret = ext4_calculate_overhead(sb);
+		if (ret)
+			goto failed_mount_wq;
+	}
+
+	/*
 	 * The maximum number of concurrent works can be high and
 	 * concurrency isn't really necessary.  Limit it to 1.
 	 */
@@ -4568,67 +4688,21 @@ restore_opts:
 	return err;
 }
 
-/*
- * Note: calculating the overhead so we can be compatible with
- * historical BSD practice is quite difficult in the face of
- * clusters/bigalloc.  This is because multiple metadata blocks from
- * different block group can end up in the same allocation cluster.
- * Calculating the exact overhead in the face of clustered allocation
- * requires either O(all block bitmaps) in memory or O(number of block
- * groups**2) in time.  We will still calculate the superblock for
- * older file systems --- and if we come across with a bigalloc file
- * system with zero in s_overhead_clusters the estimate will be close to
- * correct especially for very large cluster sizes --- but for newer
- * file systems, it's better to calculate this figure once at mkfs
- * time, and store it in the superblock.  If the superblock value is
- * present (even for non-bigalloc file systems), we will use it.
- */
 static int ext4_statfs(struct dentry *dentry, struct kstatfs *buf)
 {
 	struct super_block *sb = dentry->d_sb;
 	struct ext4_sb_info *sbi = EXT4_SB(sb);
 	struct ext4_super_block *es = sbi->s_es;
-	struct ext4_group_desc *gdp;
+	ext4_fsblk_t overhead = 0;
 	u64 fsid;
 	s64 bfree;
 
-	if (test_opt(sb, MINIX_DF)) {
-		sbi->s_overhead_last = 0;
-	} else if (es->s_overhead_clusters) {
-		sbi->s_overhead_last = le32_to_cpu(es->s_overhead_clusters);
-	} else if (sbi->s_blocks_last != ext4_blocks_count(es)) {
-		ext4_group_t i, ngroups = ext4_get_groups_count(sb);
-		ext4_fsblk_t overhead = 0;
-
-		/*
-		 * Compute the overhead (FS structures).  This is constant
-		 * for a given filesystem unless the number of block groups
-		 * changes so we cache the previous value until it does.
-		 */
-
-		/*
-		 * All of the blocks before first_data_block are
-		 * overhead
-		 */
-		overhead = EXT4_B2C(sbi, le32_to_cpu(es->s_first_data_block));
-
-		/*
-		 * Add the overhead found in each block group
-		 */
-		for (i = 0; i < ngroups; i++) {
-			gdp = ext4_get_group_desc(sb, i, NULL);
-			overhead += ext4_num_overhead_clusters(sb, i, gdp);
-			cond_resched();
-		}
-		sbi->s_overhead_last = overhead;
-		smp_wmb();
-		sbi->s_blocks_last = ext4_blocks_count(es);
-	}
+	if (!test_opt(sb, MINIX_DF))
+		overhead = sbi->s_overhead;
 
 	buf->f_type = EXT4_SUPER_MAGIC;
 	buf->f_bsize = sb->s_blocksize;
-	buf->f_blocks = (ext4_blocks_count(es) -
-			 EXT4_C2B(sbi, sbi->s_overhead_last));
+	buf->f_blocks = ext4_blocks_count(es) - EXT4_C2B(sbi, sbi->s_overhead);
 	bfree = percpu_counter_sum_positive(&sbi->s_freeclusters_counter) -
 		percpu_counter_sum_positive(&sbi->s_dirtyclusters_counter);
 	/* prevent underflow in case that few free space is available */


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/