Patchwork ext4: fix free space estimate in ext4_nonda_switch()

login
register
mail settings
Submitter Eric Whitney
Date April 3, 2013, 5:19 p.m.
Message ID <20130403171931.GC3006@wallace>
Download mbox | patch
Permalink /patch/233560/
State Accepted
Headers show

Comments

Eric Whitney - April 3, 2013, 5:19 p.m.
Values stored in s_freeclusters_counter and s_dirtyclusters_counter
are both in cluster units.  Remove the cluster to block conversion
applied to s_freeclusters_counter causing an inflated estimate of
free space because s_dirtyclusters_counter is not similarly
converted.  Rename free_blocks and dirty_blocks to better reflect
the units these variables contain to avoid future confusion.  This
fix corrects ENOSPC failures for xfstests 127 and 231 on bigalloc
file systems.

Signed-off-by: Eric Whitney <enwlinux@gmail.com>
---
 fs/ext4/inode.c |   15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)
Theodore Ts'o - April 9, 2013, 1:28 p.m.
On Wed, Apr 03, 2013 at 01:19:31PM -0400, Eric Whitney wrote:
> Values stored in s_freeclusters_counter and s_dirtyclusters_counter
> are both in cluster units.  Remove the cluster to block conversion
> applied to s_freeclusters_counter causing an inflated estimate of
> free space because s_dirtyclusters_counter is not similarly
> converted.  Rename free_blocks and dirty_blocks to better reflect
> the units these variables contain to avoid future confusion.  This
> fix corrects ENOSPC failures for xfstests 127 and 231 on bigalloc
> file systems.
> 
> Signed-off-by: Eric Whitney <enwlinux@gmail.com>

Hi Eric,

This patch looks good.  Thanks for submitting it!!

I've applied it to the dev branch for testing.

					- Ted
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index b3a5213..5f5cc40 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -2661,7 +2661,7 @@  out_writepages:
 
 static int ext4_nonda_switch(struct super_block *sb)
 {
-	s64 free_blocks, dirty_blocks;
+	s64 free_clusters, dirty_clusters;
 	struct ext4_sb_info *sbi = EXT4_SB(sb);
 
 	/*
@@ -2672,17 +2672,18 @@  static int ext4_nonda_switch(struct super_block *sb)
 	 * Delalloc need an accurate free block accounting. So switch
 	 * to non delalloc when we are near to error range.
 	 */
-	free_blocks  = EXT4_C2B(sbi,
-		percpu_counter_read_positive(&sbi->s_freeclusters_counter));
-	dirty_blocks = percpu_counter_read_positive(&sbi->s_dirtyclusters_counter);
+	free_clusters =
+		percpu_counter_read_positive(&sbi->s_freeclusters_counter);
+	dirty_clusters =
+		percpu_counter_read_positive(&sbi->s_dirtyclusters_counter);
 	/*
 	 * Start pushing delalloc when 1/2 of free blocks are dirty.
 	 */
-	if (dirty_blocks && (free_blocks < 2 * dirty_blocks))
+	if (dirty_clusters && (free_clusters < 2 * dirty_clusters))
 		try_to_writeback_inodes_sb(sb, WB_REASON_FS_FREE_SPACE);
 
-	if (2 * free_blocks < 3 * dirty_blocks ||
-		free_blocks < (dirty_blocks + EXT4_FREECLUSTERS_WATERMARK)) {
+	if (2 * free_clusters < 3 * dirty_clusters ||
+	    free_clusters < (dirty_clusters + EXT4_FREECLUSTERS_WATERMARK)) {
 		/*
 		 * free block count is less than 150% of dirty blocks
 		 * or free blocks is less than watermark