Skip to contentSkip to conversation

commit_metadata export operation v5

PATCH In review Ben Myers 4 participants 7 messages Feb 16, 2010 – Feb 17, 2010

Subsystems (4)
lore
Share & tools
lore

Ben Myers

author
#0 lore raw

cover commit_metadata export operation v5

Hey Bruce,

Here is the latest version of the knfsd sync changes that I've been spamming you with. I have addressed the latest suggestions provided by Christoph, Trond, Dave, and Alex. This version of the commit_metadata export operation takes only one inode as suggested by Christoph and it turns out to be much cleaner this way. Now we've gone back to committing all of the time in nfsd_create_setattr and don't bother with the added argument. A couple extra comments to explain the commit ordering.

Thanks,
Ben

---

Ben Myers (2):
commit_metadata export operation replacing nfsd_sync_dir xfs_export_operations.commit_metadata

 fs/nfsd/nfs4recover.c         |    4 --
 fs/nfsd/vfs.c                 |  109 ++++++++++++++++++++---------------------
 fs/xfs/linux-2.6/xfs_export.c |   24 +++++++++
 include/linux/exportfs.h      |    5 ++
 4 files changed, 83 insertions(+), 59 deletions(-)
Signature · 4 lines
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
author
#1 lore raw

PATCH 1/2 commit_metadata export operation replacing nfsd_sync_dir

- Add commit_metadata export_operation to allow the underlying filesystem to decide how to commit an inode most efficiently.

- Usage of nfsd_sync_dir and write_inode_now has been replaced with the commit_metadata function that takes a svc_fh.

- The commit_metadata function calls the commit_metadata export_op if it's there, or else falls back to sync_inode instead of fsync and write_inode_now because only metadata need be synced here.

- nfsd4_sync_rec_dir now uses vfs_fsync so that commit_metadata can be static

---
 fs/nfsd/nfs4recover.c    |    4 --
 fs/nfsd/vfs.c            |  109 ++++++++++++++++++++++------------------------
 include/linux/exportfs.h |    5 ++
 3 files changed, 59 insertions(+), 59 deletions(-)
fs/nfsd/nfs4recover.c +1 -3
index 5a754f7..98fb98e 100644
@@ -119,9 +119,7 @@ out_no_tfm:
 static void
 nfsd4_sync_rec_dir(void)
 {
-	mutex_lock(&rec_dir.dentry->d_inode->i_mutex);
-	nfsd_sync_dir(rec_dir.dentry);
-	mutex_unlock(&rec_dir.dentry->d_inode->i_mutex);
+	vfs_fsync(NULL, rec_dir.dentry, 0);
 }
 
 int
fs/nfsd/vfs.c +53 -56

185 lines not on this page · raw message

include/linux/exportfs.h +5 -0
index dc12f41..a9cd507 100644
@@ -96,6 +96,7 @@ struct fid {
  * @fh_to_parent:   find the implied object's parent and get a dentry for it
  * @get_name:       find the name for a given inode in a given directory
  * @get_parent:     find the parent of a given directory
+ * @commit_metadata: commit metadata changes to stable storage
  *
  * See Documentation/filesystems/nfs/Exporting for details on how to use
  * this interface correctly.
@@ -137,6 +138,9 @@ struct fid {
  *    is also a directory.  In the event that it cannot be found, or storage
  *    space cannot be allocated, a %ERR_PTR should be returned.
  *
+ * commit_metadata:
+ *    @commit_metadata should commit metadata changes to stable storage.
+ *
  * Locking rules:
  *    get_parent is called with child->d_inode->i_mutex down
  *    get_name is not (which is possibly inconsistent)
@@ -152,6 +156,7 @@ struct export_operations {
 	int (*get_name)(struct dentry *parent, char *name,
 			struct dentry *child);
 	struct dentry * (*get_parent)(struct dentry *child);
+	int (*commit_metadata)(struct inode *inode);
 };
 
 extern int exportfs_encode_fh(struct dentry *dentry, struct fid *fid,

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#2 lore raw

Re: [PATCH 1/2] commit_metadata export operation replacing nfsd_sync_dir

This looks very good to me. A couple of tiny nitpicks below:

Quoted 3 lines

+/*
+ * Commit metadata changes to stable storage. You pay pass NULL for dchild.
+ */

The dchild argument is gone in this version.

Quoted 4 lines
fs/nfsd/vfs.c +4 -0
+	struct writeback_control wbc = {
+		.sync_mode = WB_SYNC_ALL,
+		.nr_to_write = 0, /* metadata only */
+	};

fs/nfsd/vfs.c out: · line 297

Quoted 10 lines
fs/nfsd/vfs.c +10 -0
+	int error = 0;
+
+	if (!EX_ISSYNC(fhp->fh_export))
+		return 0;
+
+	if (export_ops->commit_metadata) {
+		error = export_ops->commit_metadata(inode);
+	} else {
+		error = sync_inode(inode, &wbc);
+	}

Maybe move the wbc declaration into the else branch here to keep variables in the smallest possible scope.

fs/nfsd/vfs.c nfsd_create_setattr(struct svc_rqst *rqstp, struct svc_fh *resfhp, · line 1207

Quoted 6 lines
fs/nfsd/vfs.c +1 -1
@@ -1199,7 +1204,7 @@ nfsd_create_setattr(struct svc_rqst *rqstp, struct svc_fh *resfhp,
 	if (current_fsuid() != 0)
 		iap->ia_valid &= ~(ATTR_UID|ATTR_GID);
 	if (iap->ia_valid)
-		return nfsd_setattr(rqstp, resfhp, iap, 0, (time_t)0);
+		return nfsd_setattr(rqstp, resfhp, iap, 0, 0);

While this is a worthwhile cleanup I'd not put it into a patch that is now entirely unrelated.

fs/nfsd/vfs.c nfsd_create(struct svc_rqst *rqstp, struct svc_fh *fhp, · line 1348

Quoted 10 lines
fs/nfsd/vfs.c +8 -0
+	err = nfsd_create_setattr(rqstp, resfhp, iap);
 
+	/*
+	 * nfsd_setattr already committed the child.  Transactional filesystems
+	 * had a chance to commit changes for both parent and child
+	 * simultaneously making the following commit_metadata a noop.
+	 */
+	err2 = nfserrno(commit_metadata(fhp));
+       	if (err2)
 		err = err2;

The if statement above seems rather minindented, possibly due to the partial use of spaces instead of tabs.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

author
#3 lore raw

PATCH 2/2 xfs_export_operations.commit_metadata

In reply to Ben Myers

This is the commit_metadata export operation for XFS.

- Takes one inode to be committed.

- Forces the log up to the lsn of the inode.

- Doesn't force the log if the inode doesn't have a pincount.

---
 fs/xfs/linux-2.6/xfs_export.c |   24 ++++++++++++++++++++++++
 1 files changed, 24 insertions(+), 0 deletions(-)
fs/xfs/linux-2.6/xfs_export.c +24 -0
index 87b8cbd..47a8d1f 100644
@@ -29,6 +29,7 @@
 #include "xfs_vnodeops.h"
 #include "xfs_bmap_btree.h"
 #include "xfs_inode.h"
+#include "xfs_inode_item.h"
 
 /*
  * Note that we only accept fileids which are long enough rather than allow
@@ -215,9 +216,32 @@ xfs_fs_get_parent(
 	return d_obtain_alias(VFS_I(cip));
 }
 
+STATIC int
+xfs_fs_nfs_commit_metadata(
+	struct inode		*inode)
+{
+	struct xfs_inode	*ip = XFS_I(inode);
+	struct xfs_mount	*mp = ip->i_mount;
+	xfs_lsn_t		force_lsn = NULLCOMMITLSN;
+	int			error = 0;
+
+	xfs_ilock(ip, XFS_ILOCK_SHARED);
+	if (xfs_ipincount(ip)) {
+		force_lsn = ip->i_itemp->ili_last_lsn;
+	}
+	if (force_lsn != NULLCOMMITLSN) {
+		error = _xfs_log_force(mp, force_lsn,
+				XFS_LOG_FORCE | XFS_LOG_SYNC, NULL);
+	}
+	xfs_iunlock(ip, XFS_ILOCK_SHARED);
+	
+	return error;
+}
+
 const struct export_operations xfs_export_operations = {
 	.encode_fh		= xfs_fs_encode_fh,
 	.fh_to_dentry		= xfs_fs_fh_to_dentry,
 	.fh_to_parent		= xfs_fs_fh_to_parent,
 	.get_parent		= xfs_fs_get_parent,
+	.commit_metadata	= xfs_fs_nfs_commit_metadata,
 };

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#5 lore raw

Re: [PATCH 2/2] xfs_export_operations.commit_metadata

In reply to Ben Myers
Quoted 36 lines · Ben Myers

This is the commit_metadata export operation for XFS.

- Takes one inode to be committed.

- Forces the log up to the lsn of the inode.

- Doesn't force the log if the inode doesn't have a pincount.

---
 fs/xfs/linux-2.6/xfs_export.c |   24 ++++++++++++++++++++++++
 1 files changed, 24 insertions(+), 0 deletions(-)
fs/xfs/linux-2.6/xfs_export.c +8 -0
index 87b8cbd..47a8d1f 100644
@@ -29,6 +29,7 @@
 #include "xfs_vnodeops.h"
 #include "xfs_bmap_btree.h"
 #include "xfs_inode.h"
+#include "xfs_inode_item.h"
 
 /*
  * Note that we only accept fileids which are long enough rather than allow
@@ -215,9 +216,32 @@ xfs_fs_get_parent(
 	return d_obtain_alias(VFS_I(cip));
 }
 
+STATIC int
+xfs_fs_nfs_commit_metadata(
+	struct inode		*inode)
+{
+	struct xfs_inode	*ip = XFS_I(inode);
+	struct xfs_mount	*mp = ip->i_mount;
+	xfs_lsn_t		force_lsn = NULLCOMMITLSN;

fs/xfs/linux-2.6/xfs_export.c xfs_fs_get_parent( · line 235

Quoted 10 lines
fs/xfs/linux-2.6/xfs_export.c +10 -0
+	int			error = 0;
+
+	xfs_ilock(ip, XFS_ILOCK_SHARED);
+	if (xfs_ipincount(ip)) {
+		force_lsn = ip->i_itemp->ili_last_lsn;
+	}
+	if (force_lsn != NULLCOMMITLSN) {
+		error = _xfs_log_force(mp, force_lsn,
+				XFS_LOG_FORCE | XFS_LOG_SYNC, NULL);
+	}

That could be simplified to:

	if (xfs_ipincount(ip))
		_xfs_log_force(mp, ip->i_itemp->ili_last_lsn
				XFS_LOG_FORCE | XFS_LOG_SYNC, NULL);

Cheers,

Dave.

Signature · 7 lines
Dave Chinner
[email protected]

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
#6 lore raw
In reply to Ben Myers
Quoted 2 lines · Ben Myers

Here is the latest version of the knfsd sync changes that I've been spamming you with.

Don't worry about spamming--it's good that people have lots of comments!

--b.

Quoted 17 lines

I have addressed the latest suggestions provided by Christoph, Trond, Dave, and Alex. This version of the commit_metadata export operation takes only one inode as suggested by Christoph and it turns out to be much cleaner this way. Now we've gone back to committing all of the time in nfsd_create_setattr and don't bother with the added argument. A couple extra comments to explain the commit ordering.

Thanks,
Ben

---

Ben Myers (2):
commit_metadata export operation replacing nfsd_sync_dir xfs_export_operations.commit_metadata

Quoted 7 lines
 fs/nfsd/nfs4recover.c         |    4 --
 fs/nfsd/vfs.c                 |  109 ++++++++++++++++++++---------------------
 fs/xfs/linux-2.6/xfs_export.c |   24 +++++++++
 include/linux/exportfs.h      |    5 ++
 4 files changed, 83 insertions(+), 59 deletions(-)

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs