add a vfs_fsync helper

Fsync currently has a fdatawrite/fdatawait pair around the method call, and a mutex_lock/unlock of the inode mutex. All callers of fsync have to duplicate this, but we have a few and most of them don't quite get it right. This patch adds a new vfs_fsync that takes care of this. It's a little more complicated as usual as ->fsync might get a NULL file pointer and just a dentry from nfsd, but otherwise gets afile and we want to take the mapping and file operations from it when it is there.

Notes on the fsync callers:

- ecryptfs wasn't calling filemap_fdatawrite / filemap_fdatawait on the lower file
- coda wasn't calling filemap_fdatawrite / filemap_fdatawait on the host

	file, and returning 0 when ->fsync was missing

- shm wasn't calling either filemap_fdatawrite / filemap_fdatawait nor taking i_mutex. Now given that shared memory doesn't have disk backing not doing anything in fsync seems fine and I left it out of the vfs_fsync conversion for now, but in that case we might just not pass it through to the lower file at all but just call the no-op simple_sync_file directly.

Seems like an improvement.. I've wanted to know, though, what do we need i_mutex for? Is it just convention, or is there some good reason to have it in generic code?

Quoted 3 lines · Nick Piggin

Seems like an improvement.. I've wanted to know, though, what do we need i_mutex for? Is it just convention, or is there some good reason to have it in generic code?

At least XFS doesn't need it. Same for the filemap_fdatawrite / filemap_fdatawait which at least for filesystems that want to provide integrity guarantees is in the wrong place.

This patch is a first one out my work to refactor fsync, and I'm trying to feed it to mainline in pieces. The next one will be to make sure

nfsd always has a struct file available when calling fsync, but I need to do some extensive benchmarking. After that we can change the fsync prototype to drop the dentry paramters, and move the filemap_fdatawrite / filemap_fdatawait aswell as the i_mutex locking into the actual methods.

Quoted 3 lines · Nick Piggin

Seems like an improvement.. I've wanted to know, though, what do we need i_mutex for? Is it just convention, or is there some good reason to have it in generic code?

At least XFS doesn't need it. Same for the filemap_fdatawrite / filemap_fdatawait which at least for filesystems that want to provide integrity guarantees is in the wrong place.

This patch is a first one out my work to refactor fsync, and I'm trying to feed it to mainline in pieces. The next one will be to make sure nfsd always has a struct file available when calling fsync, but I need to do some extensive benchmarking. After that we can change the fsync prototype to drop the dentry paramters, and move the filemap_fdatawrite / filemap_fdatawait aswell as the i_mutex locking into the actual methods.

Fsync currently has a fdatawrite/fdatawait pair around the method call, and a mutex_lock/unlock of the inode mutex. All callers of fsync have to duplicate this, but we have a few and most of them don't quite get it right. This patch adds a new vfs_fsync that takes care of this. It's a little more complicated as usual as ->fsync might get a NULL file pointer and just a dentry from nfsd, but otherwise gets afile and we want to take the mapping and file operations from it when it is there.

Notes on the fsync callers:

- ecryptfs wasn't calling filemap_fdatawrite / filemap_fdatawait on the lower file
- coda wasn't calling filemap_fdatawrite / filemap_fdatawait on the host

	file, and returning 0 when ->fsync was missing

- shm wasn't calling either filemap_fdatawrite / filemap_fdatawait nor taking i_mutex. Now given that shared memory doesn't have disk backing not doing anything in fsync seems fine and I left it out of the vfs_fsync conversion for now, but in that case we might just not pass it through to the lower file at all but just call the no-op simple_sync_file directly.

[and now actually export vfs_fsync]

Signed-off-by Christoph Hellwig <[email protected]>

Index: xfs/fs/nfsd/vfs.c
===================================================================

fs/nfsd/vfs.c +3 -32

@@ -743,45 +743,16 @@ nfsd_close(struct file *filp)
 	fput(filp);
 }
 
-/*
- * Sync a file
- * As this calls fsync (not fdatasync) there is no need for a write_inode
- * after it.
- */
-static inline int nfsd_dosync(struct file *filp, struct dentry *dp,
-			      const struct file_operations *fop)
-{
-	struct inode *inode = dp->d_inode;
-	int (*fsync) (struct file *, struct dentry *, int);
-	int err;
-
-	err = filemap_fdatawrite(inode->i_mapping);
-	if (err == 0 && fop && (fsync = fop->fsync))
-		err = fsync(filp, dp, 0);
-	if (err == 0)
-		err = filemap_fdatawait(inode->i_mapping);
-
-	return err;
-}
-	
-
 static int
 nfsd_sync(struct file *filp)
 {
-        int err;
-	struct inode *inode = filp->f_path.dentry->d_inode;
-	dprintk("nfsd: sync file %s\n", filp->f_path.dentry->d_name.name);
-	mutex_lock(&inode->i_mutex);
-	err=nfsd_dosync(filp, filp->f_path.dentry, filp->f_op);
-	mutex_unlock(&inode->i_mutex);
-
-	return err;
+	return vfs_fsync(filp, filp->f_path.dentry, 0);
 }
 
 int
-nfsd_sync_dir(struct dentry *dp)
+nfsd_sync_dir(struct dentry *dentry)
 {
-	return nfsd_dosync(NULL, dp, dp->d_inode->i_fop);
+	return vfs_fsync(NULL, dentry, 0);
 }
 
 /*

Index: xfs/fs/sync.c
===================================================================

fs/sync.c +37 -11

@@ -75,14 +75,39 @@ int file_fsync(struct file *filp, struct
 	return ret;
 }
 
-long do_fsync(struct file *file, int datasync)
+/**
+ * vfs_fsync - perform a fsync or fdatasync on a file
+ * @file:		file to sync
+ * @dentry:		dentry of @file
+ * @data:		only perform a fdatasync operation
+ *
+ * Write back data and metadata for @file to disk.  If @datasync is
+ * set only metadata needed to access modified file data is written.
+ *
+ * In case this function is called from nfsd @file may be %NULL and
+ * only @dentry is set.  This can only happen when the filesystem
+ * implements the export_operations API.
+ */
+int vfs_fsync(struct file *file, struct dentry *dentry, int datasync)
 {
-	int ret;
-	int err;
-	struct address_space *mapping = file->f_mapping;
+	const struct file_operations *fop;
+	struct address_space *mapping;
+	int err, ret;
+
+	/*
+	 * Get mapping and operations from the file in case we have
+	 * as file, or get the default values for them in case we
+	 * don't have a struct file available.  Damn nfsd..
+	 */
+	if (file) {
+		mapping = file->f_mapping;
+		fop = file->f_op;
+	} else {
+		mapping = dentry->d_inode->i_mapping;
+		fop = dentry->d_inode->i_fop;
+	}
 
-	if (!file->f_op || !file->f_op->fsync) {
-		/* Why?  We can still call filemap_fdatawrite */
+	if (!fop || !fop->fsync) {
 		ret = -EINVAL;
 		goto out;
 	}

@@ -94,7 +119,7 @@ long do_fsync(struct file *file, int dat
 	 * livelocks in fsync_buffers_list().
 	 */
 	mutex_lock(&mapping->host->i_mutex);
-	err = file->f_op->fsync(file, file->f_path.dentry, datasync);
+	err = fop->fsync(file, dentry, datasync);
 	if (!ret)
 		ret = err;
 	mutex_unlock(&mapping->host->i_mutex);

@@ -104,15 +129,16 @@ long do_fsync(struct file *file, int dat
 out:
 	return ret;
 }
+EXPORT_SYMBOL(vfs_fsync);
 
-static long __do_fsync(unsigned int fd, int datasync)
+static int do_fsync(unsigned int fd, int datasync)
 {
 	struct file *file;
 	int ret = -EBADF;
 
 	file = fget(fd);
 	if (file) {
-		ret = do_fsync(file, datasync);
+		ret = vfs_fsync(file, file->f_path.dentry, datasync);
 		fput(file);
 	}
 	return ret;

@@ -120,12 +146,12 @@ static long __do_fsync(unsigned int fd, 
 
 asmlinkage long sys_fsync(unsigned int fd)
 {
-	return __do_fsync(fd, 0);
+	return do_fsync(fd, 0);
 }
 
 asmlinkage long sys_fdatasync(unsigned int fd)
 {
-	return __do_fsync(fd, 1);
+	return do_fsync(fd, 1);
 }
 
 /*

Index: xfs/include/linux/fs.h
===================================================================

include/linux/fs.h +1 -1

@@ -1827,7 +1827,7 @@ extern int __filemap_fdatawrite_range(st
 extern int filemap_fdatawrite_range(struct address_space *mapping,
 				loff_t start, loff_t end);
 
-extern long do_fsync(struct file *file, int datasync);
+extern int vfs_fsync(struct file *file, struct dentry *dentry, int datasync);
 extern void sync_supers(void);
 extern void sync_filesystems(int wait);
 extern void __fsync_super(struct super_block *sb);

Index: xfs/mm/msync.c
===================================================================

mm/msync.c +1 -1

@@ -82,7 +82,7 @@ asmlinkage long sys_msync(unsigned long 
 				(vma->vm_flags & VM_SHARED)) {
 			get_file(file);
 			up_read(&mm->mmap_sem);
-			error = do_fsync(file, 0);
+			error = vfs_fsync(file, file->f_path.dentry, 0);
 			fput(file);
 			if (error || start >= end)
 				goto out;

Index: xfs/drivers/usb/gadget/file_storage.c
===================================================================

drivers/usb/gadget/file_storage.c +1 -17

@@ -1863,26 +1863,10 @@ static int do_write(struct fsg_dev *fsg)
 static int fsync_sub(struct lun *curlun)
 {
 	struct file	*filp = curlun->filp;
-	struct inode	*inode;
-	int		rc, err;
 
 	if (curlun->ro || !filp)
 		return 0;
-	if (!filp->f_op->fsync)
-		return -EINVAL;
-
-	inode = filp->f_path.dentry->d_inode;
-	mutex_lock(&inode->i_mutex);
-	rc = filemap_fdatawrite(inode->i_mapping);
-	err = filp->f_op->fsync(filp, filp->f_path.dentry, 1);
-	if (!rc)
-		rc = err;
-	err = filemap_fdatawait(inode->i_mapping);
-	if (!rc)
-		rc = err;
-	mutex_unlock(&inode->i_mutex);
-	VLDBG(curlun, "fdatasync -> %d\n", rc);
-	return rc;
+	return vfs_fsync(filp, filp->f_path.dentry, 1);
 }
 
 static void fsync_all(struct fsg_dev *fsg)

Index: xfs/fs/coda/file.c
===================================================================

fs/coda/file.c +2 -10

@@ -200,8 +200,7 @@ int coda_release(struct inode *coda_inod
 int coda_fsync(struct file *coda_file, struct dentry *coda_dentry, int datasync)
 {
 	struct file *host_file;
-	struct dentry *host_dentry;
-	struct inode *host_inode, *coda_inode = coda_dentry->d_inode;
+	struct inode *coda_inode = coda_dentry->d_inode;
 	struct coda_file_info *cfi;
 	int err = 0;
 

@@ -213,14 +212,7 @@ int coda_fsync(struct file *coda_file, s
 	BUG_ON(!cfi || cfi->cfi_magic != CODA_MAGIC);
 	host_file = cfi->cfi_container;
 
-	if (host_file->f_op && host_file->f_op->fsync) {
-		host_dentry = host_file->f_path.dentry;
-		host_inode = host_dentry->d_inode;
-		mutex_lock(&host_inode->i_mutex);
-		err = host_file->f_op->fsync(host_file, host_dentry, datasync);
-		mutex_unlock(&host_inode->i_mutex);
-	}
-
+	err = vfs_fsync(host_file, host_file->f_path.dentry, datasync);
 	if ( !err && !datasync ) {
 		lock_kernel();
 		err = venus_fsync(coda_inode->i_sb, coda_i2f(coda_inode));

Index: xfs/fs/ecryptfs/file.c
===================================================================

fs/ecryptfs/file.c +3 -12

@@ -275,18 +275,9 @@ static int ecryptfs_release(struct inode
 static int
 ecryptfs_fsync(struct file *file, struct dentry *dentry, int datasync)
 {
-	struct file *lower_file = ecryptfs_file_to_lower(file);
-	struct dentry *lower_dentry = ecryptfs_dentry_to_lower(dentry);
-	struct inode *lower_inode = lower_dentry->d_inode;
-	int rc = -EINVAL;
-
-	if (lower_inode->i_fop->fsync) {
-		mutex_lock(&lower_inode->i_mutex);
-		rc = lower_inode->i_fop->fsync(lower_file, lower_dentry,
-					       datasync);
-		mutex_unlock(&lower_inode->i_mutex);
-	}
-	return rc;
+	return vfs_fsync(ecryptfs_file_to_lower(file),
+			 ecryptfs_dentry_to_lower(dentry),
+			 datasync);
 }
 
 static int ecryptfs_fasync(int fd, struct file *file, int flag)