2011-04-05 23:24:10

by Curt Wohlgemuth

[permalink] [raw]
Subject: [PATCH] ext4: sync the directory inode in ext4_sync_parent()

ext4 has taken the stance that, in the absence of a journal,
when an fsync/fdatasync of an inode is done, the parent
directory should be sync'ed if this inode entry is new.
ext4_sync_parent(), which implements this, does indeed sync
the dirent pages for parent directories, but it does not
sync the directory *inode*. This patch fixes this.

Also now return error status from ext4_sync_parent().

I tested this using a power fail test, which panics a
machine running a file server getting requests from a
client. Without this patch, on about every other test run,
the server is missing many, many files that had been synced.
With this patch, on > 6 runs, I see zero files being lost.

Signed-off-by: Curt Wohlgemuth <[email protected]>
---
fs/ext4/fsync.c | 18 +++++++++++++++---
1 files changed, 15 insertions(+), 3 deletions(-)

diff --git a/fs/ext4/fsync.c b/fs/ext4/fsync.c
index 7f74019..bb0a4e7 100644
--- a/fs/ext4/fsync.c
+++ b/fs/ext4/fsync.c
@@ -125,9 +125,11 @@ extern int ext4_flush_completed_IO(struct inode *inode)
* the parent directory's parent as well, and so on recursively, if
* they are also freshly created.
*/
-static void ext4_sync_parent(struct inode *inode)
+static int ext4_sync_parent(struct inode *inode)
{
struct dentry *dentry = NULL;
+ int ret = 0;
+ struct writeback_control wbc;

while (inode && ext4_test_inode_state(inode, EXT4_STATE_NEWENTRY)) {
ext4_clear_inode_state(inode, EXT4_STATE_NEWENTRY);
@@ -136,8 +138,18 @@ static void ext4_sync_parent(struct inode *inode)
if (!dentry || !dentry->d_parent || !dentry->d_parent->d_inode)
break;
inode = dentry->d_parent->d_inode;
- sync_mapping_buffers(inode->i_mapping);
+ ret = sync_mapping_buffers(inode->i_mapping);
+ if (ret)
+ goto out;
+ memset(&wbc, 0, sizeof(wbc));
+ wbc.sync_mode = WB_SYNC_ALL;
+ wbc.nr_to_write = 0; /* only write out the inode */
+ ret = sync_inode(inode, &wbc);
+ if (ret)
+ goto out;
}
+out:
+ return ret;
}

/*
@@ -176,7 +188,7 @@ int ext4_sync_file(struct file *file, int datasync)
if (!journal) {
ret = generic_file_fsync(file, datasync);
if (!ret && !list_empty(&inode->i_dentry))
- ext4_sync_parent(inode);
+ ret = ext4_sync_parent(inode);
goto out;
}

--
1.7.3.1



2011-04-06 21:01:41

by Theodore Ts'o

[permalink] [raw]
Subject: Re: [PATCH] ext4: sync the directory inode in ext4_sync_parent()

Hi Curt,

Thanks, accepted into the tree. I made a few tiny stylistic changes
(mainly by replacing "goto out" with "break").

- Ted

>From 8429a2e121372395f25d80a486ef065e77139ebb Mon Sep 17 00:00:00 2001
From: Curt Wohlgemuth <[email protected]>
Date: Wed, 6 Apr 2011 16:57:09 -0400
Subject: [PATCH] ext4: sync the directory inode in ext4_sync_parent()

ext4 has taken the stance that, in the absence of a journal,
when an fsync/fdatasync of an inode is done, the parent
directory should be sync'ed if this inode entry is new.
ext4_sync_parent(), which implements this, does indeed sync
the dirent pages for parent directories, but it does not
sync the directory *inode*. This patch fixes this.

Also now return error status from ext4_sync_parent().

I tested this using a power fail test, which panics a
machine running a file server getting requests from a
client. Without this patch, on about every other test run,
the server is missing many, many files that had been synced.
With this patch, on > 6 runs, I see zero files being lost.

Signed-off-by: Curt Wohlgemuth <[email protected]>
Signed-off-by: "Theodore Ts'o" <[email protected]>
---
fs/ext4/fsync.c | 17 ++++++++++++++---
1 files changed, 14 insertions(+), 3 deletions(-)

diff --git a/fs/ext4/fsync.c b/fs/ext4/fsync.c
index 7f74019..b1f9b5f 100644
--- a/fs/ext4/fsync.c
+++ b/fs/ext4/fsync.c
@@ -125,9 +125,11 @@ extern int ext4_flush_completed_IO(struct inode *inode)
* the parent directory's parent as well, and so on recursively, if
* they are also freshly created.
*/
-static void ext4_sync_parent(struct inode *inode)
+static int ext4_sync_parent(struct inode *inode)
{
+ struct writeback_control wbc;
struct dentry *dentry = NULL;
+ int ret = 0;

while (inode && ext4_test_inode_state(inode, EXT4_STATE_NEWENTRY)) {
ext4_clear_inode_state(inode, EXT4_STATE_NEWENTRY);
@@ -136,8 +138,17 @@ static void ext4_sync_parent(struct inode *inode)
if (!dentry || !dentry->d_parent || !dentry->d_parent->d_inode)
break;
inode = dentry->d_parent->d_inode;
- sync_mapping_buffers(inode->i_mapping);
+ ret = sync_mapping_buffers(inode->i_mapping);
+ if (ret)
+ break;
+ memset(&wbc, 0, sizeof(wbc));
+ wbc.sync_mode = WB_SYNC_ALL;
+ wbc.nr_to_write = 0; /* only write out the inode */
+ ret = sync_inode(inode, &wbc);
+ if (ret)
+ break;
}
+ return ret;
}

/*
@@ -176,7 +187,7 @@ int ext4_sync_file(struct file *file, int datasync)
if (!journal) {
ret = generic_file_fsync(file, datasync);
if (!ret && !list_empty(&inode->i_dentry))
- ext4_sync_parent(inode);
+ ret = ext4_sync_parent(inode);
goto out;
}

--
1.7.3.1