cleanup block device inode syncing

cleanup block device inode syncing

[Christoph Hellwig]

Hi Jens,

this series refactors parts of the sync code so that we have and always
use proper helpers for syncing data cached in the block device inode.

Diffstat:
 block/bdev.c                       |   28 +++++++++++-----
 drivers/block/xen-blkback/xenbus.c |    2 -
 fs/btrfs/volumes.c                 |    2 -
 fs/fat/inode.c                     |    6 +--
 fs/internal.h                      |   11 ------
 fs/ntfs3/inode.c                   |    2 -
 fs/sync.c                          |   62 +++++++++++++------------------------
 include/linux/blkdev.h             |    9 +++++
 8 files changed, 56 insertions(+), 66 deletions(-)

[PATCH 1/7] fs: remove __sync_filesystem

[Christoph Hellwig]

There is no clear benefit in having this helper vs just open coding it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 fs/sync.c | 38 +++++++++++++++++---------------------
 1 file changed, 17 insertions(+), 21 deletions(-)

diff --git a/fs/sync.c b/fs/sync.c
index 1373a610dc784..0d6cdc507cb98 100644
--- a/fs/sync.c
+++ b/fs/sync.c
@@ -21,25 +21,6 @@
 #define VALID_FLAGS (SYNC_FILE_RANGE_WAIT_BEFORE|SYNC_FILE_RANGE_WRITE| \
 			SYNC_FILE_RANGE_WAIT_AFTER)
 
-/*
- * Do the filesystem syncing work. For simple filesystems
- * writeback_inodes_sb(sb) just dirties buffers with inodes so we have to
- * submit IO for these buffers via __sync_blockdev(). This also speeds up the
- * wait == 1 case since in that case write_inode() functions do
- * sync_dirty_buffer() and thus effectively write one block at a time.
- */
-static int __sync_filesystem(struct super_block *sb, int wait)
-{
-	if (wait)
-		sync_inodes_sb(sb);
-	else
-		writeback_inodes_sb(sb, WB_REASON_SYNC);
-
-	if (sb->s_op->sync_fs)
-		sb->s_op->sync_fs(sb, wait);
-	return __sync_blockdev(sb->s_bdev, wait);
-}
-
 /*
  * Write out and wait upon all dirty data associated with this
  * superblock.  Filesystem data as well as the underlying block
@@ -61,10 +42,25 @@ int sync_filesystem(struct super_block *sb)
 	if (sb_rdonly(sb))
 		return 0;
 
-	ret = __sync_filesystem(sb, 0);
+	/*
+	 * Do the filesystem syncing work.  For simple filesystems
+	 * writeback_inodes_sb(sb) just dirties buffers with inodes so we have
+	 * to submit I/O for these buffers via __sync_blockdev().  This also
+	 * speeds up the wait == 1 case since in that case write_inode()
+	 * methods call sync_dirty_buffer() and thus effectively write one block
+	 * at a time.
+	 */
+	writeback_inodes_sb(sb, WB_REASON_SYNC);
+	if (sb->s_op->sync_fs)
+		sb->s_op->sync_fs(sb, 0);
+	ret = __sync_blockdev(sb->s_bdev, 0);
 	if (ret < 0)
 		return ret;
-	return __sync_filesystem(sb, 1);
+
+	sync_inodes_sb(sb);
+	if (sb->s_op->sync_fs)
+		sb->s_op->sync_fs(sb, 1);
+	return __sync_blockdev(sb->s_bdev, 1);
 }
 EXPORT_SYMBOL(sync_filesystem);
 
-- 
2.30.2

Re: [PATCH 1/7] fs: remove __sync_filesystem

[Chaitanya Kulkarni]

On 10/18/2021 11:25 PM, Christoph Hellwig wrote:
> There is no clear benefit in having this helper vs just open coding it.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>

Especially if there is only one caller.

Looks good.

Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>

[PATCH 2/7] block: remove __sync_blockdev

[Christoph Hellwig]

Instead offer a new sync_blockdev_nowait helper for the !wait case.
This new helper is exported as it will grow modular callers in a bit.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 block/bdev.c           | 11 ++++++-----
 fs/internal.h          |  5 -----
 fs/sync.c              |  7 ++++---
 include/linux/blkdev.h |  5 +++++
 4 files changed, 15 insertions(+), 13 deletions(-)

diff --git a/block/bdev.c b/block/bdev.c
index cff0bb3a4578f..fe91209881730 100644
--- a/block/bdev.c
+++ b/block/bdev.c
@@ -185,14 +185,13 @@ int sb_min_blocksize(struct super_block *sb, int size)
 
 EXPORT_SYMBOL(sb_min_blocksize);
 
-int __sync_blockdev(struct block_device *bdev, int wait)
+int sync_blockdev_nowait(struct block_device *bdev)
 {
 	if (!bdev)
 		return 0;
-	if (!wait)
-		return filemap_flush(bdev->bd_inode->i_mapping);
-	return filemap_write_and_wait(bdev->bd_inode->i_mapping);
+	return filemap_flush(bdev->bd_inode->i_mapping);
 }
+EXPORT_SYMBOL_GPL(sync_blockdev_nowait);
 
 /*
  * Write out and wait upon all the dirty data associated with a block
@@ -200,7 +199,9 @@ int __sync_blockdev(struct block_device *bdev, int wait)
  */
 int sync_blockdev(struct block_device *bdev)
 {
-	return __sync_blockdev(bdev, 1);
+	if (!bdev)
+		return 0;
+	return filemap_write_and_wait(bdev->bd_inode->i_mapping);
 }
 EXPORT_SYMBOL(sync_blockdev);
 
diff --git a/fs/internal.h b/fs/internal.h
index 3cd065c8a66b4..b5caa16f4645d 100644
--- a/fs/internal.h
+++ b/fs/internal.h
@@ -23,7 +23,6 @@ struct pipe_inode_info;
 #ifdef CONFIG_BLOCK
 extern void __init bdev_cache_init(void);
 
-extern int __sync_blockdev(struct block_device *bdev, int wait);
 void iterate_bdevs(void (*)(struct block_device *, void *), void *);
 void emergency_thaw_bdev(struct super_block *sb);
 #else
@@ -31,10 +30,6 @@ static inline void bdev_cache_init(void)
 {
 }
 
-static inline int __sync_blockdev(struct block_device *bdev, int wait)
-{
-	return 0;
-}
 static inline void iterate_bdevs(void (*f)(struct block_device *, void *),
 		void *arg)
 {
diff --git a/fs/sync.c b/fs/sync.c
index 0d6cdc507cb98..a621089eb07ee 100644
--- a/fs/sync.c
+++ b/fs/sync.c
@@ -3,6 +3,7 @@
  * High-level sync()-related operations
  */
 
+#include <linux/blkdev.h>
 #include <linux/kernel.h>
 #include <linux/file.h>
 #include <linux/fs.h>
@@ -45,7 +46,7 @@ int sync_filesystem(struct super_block *sb)
 	/*
 	 * Do the filesystem syncing work.  For simple filesystems
 	 * writeback_inodes_sb(sb) just dirties buffers with inodes so we have
-	 * to submit I/O for these buffers via __sync_blockdev().  This also
+	 * to submit I/O for these buffers via sync_blockdev().  This also
 	 * speeds up the wait == 1 case since in that case write_inode()
 	 * methods call sync_dirty_buffer() and thus effectively write one block
 	 * at a time.
@@ -53,14 +54,14 @@ int sync_filesystem(struct super_block *sb)
 	writeback_inodes_sb(sb, WB_REASON_SYNC);
 	if (sb->s_op->sync_fs)
 		sb->s_op->sync_fs(sb, 0);
-	ret = __sync_blockdev(sb->s_bdev, 0);
+	ret = sync_blockdev_nowait(sb->s_bdev);
 	if (ret < 0)
 		return ret;
 
 	sync_inodes_sb(sb);
 	if (sb->s_op->sync_fs)
 		sb->s_op->sync_fs(sb, 1);
-	return __sync_blockdev(sb->s_bdev, 1);
+	return sync_blockdev(sb->s_bdev);
 }
 EXPORT_SYMBOL(sync_filesystem);
 
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index fd9771a1da096..67a3b9e04233f 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -1285,6 +1285,7 @@ int truncate_bdev_range(struct block_device *bdev, fmode_t mode, loff_t lstart,
 #ifdef CONFIG_BLOCK
 void invalidate_bdev(struct block_device *bdev);
 int sync_blockdev(struct block_device *bdev);
+int sync_blockdev_nowait(struct block_device *bdev);
 #else
 static inline void invalidate_bdev(struct block_device *bdev)
 {
@@ -1293,6 +1294,10 @@ static inline int sync_blockdev(struct block_device *bdev)
 {
 	return 0;
 }
+static inline int sync_blockdev_nowait(struct block_device *bdev)
+{
+	return 0;
+}
 #endif
 int fsync_bdev(struct block_device *bdev);
 
-- 
2.30.2

[PATCH 3/7] xen-blkback: use sync_blockdev

[Christoph Hellwig]

Use sync_blockdev instead of opencoding it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 drivers/block/xen-blkback/xenbus.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/block/xen-blkback/xenbus.c b/drivers/block/xen-blkback/xenbus.c
index 33eba3df4dd9a..914587aabca0c 100644
--- a/drivers/block/xen-blkback/xenbus.c
+++ b/drivers/block/xen-blkback/xenbus.c
@@ -98,7 +98,7 @@ static void xen_update_blkif_status(struct xen_blkif *blkif)
 		return;
 	}
 
-	err = filemap_write_and_wait(blkif->vbd.bdev->bd_inode->i_mapping);
+	err = sync_blockdev(blkif->vbd.bdev);
 	if (err) {
 		xenbus_dev_error(blkif->be->dev, err, "block flush");
 		return;
-- 
2.30.2

Re: [PATCH 3/7] xen-blkback: use sync_blockdev

[Chaitanya Kulkarni]

On 10/18/2021 11:25 PM, Christoph Hellwig wrote:
> Use sync_blockdev instead of opencoding it.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> ---


Looks good.

Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>

[PATCH 4/7] btrfs: use sync_blockdev

[Christoph Hellwig]

Use sync_blockdev instead of opencoding it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 fs/btrfs/volumes.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 2ec3b8ac8fa35..b51e4b464103e 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -508,7 +508,7 @@ btrfs_get_bdev_and_sb(const char *device_path, fmode_t flags, void *holder,
 	}
 
 	if (flush)
-		filemap_write_and_wait((*bdev)->bd_inode->i_mapping);
+		sync_blockdev(*bdev);
 	ret = set_blocksize(*bdev, BTRFS_BDEV_BLOCKSIZE);
 	if (ret) {
 		blkdev_put(*bdev, flags);
-- 
2.30.2

Re: [PATCH 4/7] btrfs: use sync_blockdev

[Chaitanya Kulkarni]

On 10/18/2021 11:25 PM, Christoph Hellwig wrote:
> Use sync_blockdev instead of opencoding it.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>


Looks good.

Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>

Re: [PATCH 4/7] btrfs: use sync_blockdev

[David Sterba]

On Tue, Oct 19, 2021 at 08:25:27AM +0200, Christoph Hellwig wrote:
> Use sync_blockdev instead of opencoding it.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>

Acked-by: David Sterba <dsterba@suse.com>

[PATCH 5/7] fat: use sync_blockdev_nowait

[Christoph Hellwig]

Use sync_blockdev_nowait instead of opencoding it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 fs/fat/inode.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/fs/fat/inode.c b/fs/fat/inode.c
index de0c9b013a851..2fd5bfddb6958 100644
--- a/fs/fat/inode.c
+++ b/fs/fat/inode.c
@@ -1943,10 +1943,8 @@ int fat_flush_inodes(struct super_block *sb, struct inode *i1, struct inode *i2)
 		ret = writeback_inode(i1);
 	if (!ret && i2)
 		ret = writeback_inode(i2);
-	if (!ret) {
-		struct address_space *mapping = sb->s_bdev->bd_inode->i_mapping;
-		ret = filemap_flush(mapping);
-	}
+	if (!ret)
+		ret = sync_blockdev_nowait(sb->s_bdev);
 	return ret;
 }
 EXPORT_SYMBOL_GPL(fat_flush_inodes);
-- 
2.30.2

Re: [PATCH 5/7] fat: use sync_blockdev_nowait

[Chaitanya Kulkarni]

On 10/18/2021 11:25 PM, Christoph Hellwig wrote:
> Use sync_blockdev_nowait instead of opencoding it.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> ---

Looks good.

Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>

[PATCH 6/7] ntfs3: use sync_blockdev_nowait

[Christoph Hellwig]

Use sync_blockdev_nowait instead of opencoding it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 fs/ntfs3/inode.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/ntfs3/inode.c b/fs/ntfs3/inode.c
index 859951d785cb2..a87ab3ad3cd38 100644
--- a/fs/ntfs3/inode.c
+++ b/fs/ntfs3/inode.c
@@ -1046,7 +1046,7 @@ int ntfs_flush_inodes(struct super_block *sb, struct inode *i1,
 	if (!ret && i2)
 		ret = writeback_inode(i2);
 	if (!ret)
-		ret = filemap_flush(sb->s_bdev->bd_inode->i_mapping);
+		ret = sync_blockdev_nowait(sb->s_bdev);
 	return ret;
 }
 
-- 
2.30.2

Re: [PATCH 6/7] ntfs3: use sync_blockdev_nowait

[Chaitanya Kulkarni]

On 10/18/2021 11:25 PM, Christoph Hellwig wrote:
> Use sync_blockdev_nowait instead of opencoding it.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>


Looks good.

Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>

[PATCH 7/7] block: simplify the block device syncing code

[Christoph Hellwig]

Get rid of the indirections and just provide a sync_bdevs
helper for the generic sync code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 block/bdev.c           | 17 ++++++++++++++---
 fs/internal.h          |  6 ------
 fs/sync.c              | 23 ++++-------------------
 include/linux/blkdev.h |  4 ++++
 4 files changed, 22 insertions(+), 28 deletions(-)

diff --git a/block/bdev.c b/block/bdev.c
index fe91209881730..3a8a753f970bb 100644
--- a/block/bdev.c
+++ b/block/bdev.c
@@ -1019,7 +1019,7 @@ int __invalidate_device(struct block_device *bdev, bool kill_dirty)
 }
 EXPORT_SYMBOL(__invalidate_device);
 
-void iterate_bdevs(void (*func)(struct block_device *, void *), void *arg)
+void sync_bdevs(bool wait)
 {
 	struct inode *inode, *old_inode = NULL;
 
@@ -1050,8 +1050,19 @@ void iterate_bdevs(void (*func)(struct block_device *, void *), void *arg)
 		bdev = I_BDEV(inode);
 
 		mutex_lock(&bdev->bd_disk->open_mutex);
-		if (bdev->bd_openers)
-			func(bdev, arg);
+		if (!bdev->bd_openers) {
+			; /* skip */
+		} else if (wait) {
+			/*
+			 * We keep the error status of individual mapping so
+			 * that applications can catch the writeback error using
+			 * fsync(2). See filemap_fdatawait_keep_errors() for
+			 * details.
+			 */
+			filemap_fdatawait_keep_errors(inode->i_mapping);
+		} else {
+			filemap_fdatawrite(inode->i_mapping);
+		}
 		mutex_unlock(&bdev->bd_disk->open_mutex);
 
 		spin_lock(&blockdev_superblock->s_inode_list_lock);
diff --git a/fs/internal.h b/fs/internal.h
index b5caa16f4645d..cdd83d4899bb3 100644
--- a/fs/internal.h
+++ b/fs/internal.h
@@ -23,17 +23,11 @@ struct pipe_inode_info;
 #ifdef CONFIG_BLOCK
 extern void __init bdev_cache_init(void);
 
-void iterate_bdevs(void (*)(struct block_device *, void *), void *);
 void emergency_thaw_bdev(struct super_block *sb);
 #else
 static inline void bdev_cache_init(void)
 {
 }
-
-static inline void iterate_bdevs(void (*f)(struct block_device *, void *),
-		void *arg)
-{
-}
 static inline int emergency_thaw_bdev(struct super_block *sb)
 {
 	return 0;
diff --git a/fs/sync.c b/fs/sync.c
index a621089eb07ee..3ce8e2137f310 100644
--- a/fs/sync.c
+++ b/fs/sync.c
@@ -78,21 +78,6 @@ static void sync_fs_one_sb(struct super_block *sb, void *arg)
 		sb->s_op->sync_fs(sb, *(int *)arg);
 }
 
-static void fdatawrite_one_bdev(struct block_device *bdev, void *arg)
-{
-	filemap_fdatawrite(bdev->bd_inode->i_mapping);
-}
-
-static void fdatawait_one_bdev(struct block_device *bdev, void *arg)
-{
-	/*
-	 * We keep the error status of individual mapping so that
-	 * applications can catch the writeback error using fsync(2).
-	 * See filemap_fdatawait_keep_errors() for details.
-	 */
-	filemap_fdatawait_keep_errors(bdev->bd_inode->i_mapping);
-}
-
 /*
  * Sync everything. We start by waking flusher threads so that most of
  * writeback runs on all devices in parallel. Then we sync all inodes reliably
@@ -111,8 +96,8 @@ void ksys_sync(void)
 	iterate_supers(sync_inodes_one_sb, NULL);
 	iterate_supers(sync_fs_one_sb, &nowait);
 	iterate_supers(sync_fs_one_sb, &wait);
-	iterate_bdevs(fdatawrite_one_bdev, NULL);
-	iterate_bdevs(fdatawait_one_bdev, NULL);
+	sync_bdevs(false);
+	sync_bdevs(true);
 	if (unlikely(laptop_mode))
 		laptop_sync_completion();
 }
@@ -133,10 +118,10 @@ static void do_sync_work(struct work_struct *work)
 	 */
 	iterate_supers(sync_inodes_one_sb, &nowait);
 	iterate_supers(sync_fs_one_sb, &nowait);
-	iterate_bdevs(fdatawrite_one_bdev, NULL);
+	sync_bdevs(false);
 	iterate_supers(sync_inodes_one_sb, &nowait);
 	iterate_supers(sync_fs_one_sb, &nowait);
-	iterate_bdevs(fdatawrite_one_bdev, NULL);
+	sync_bdevs(false);
 	printk("Emergency Sync complete\n");
 	kfree(work);
 }
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 67a3b9e04233f..4bb9c621e0ac3 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -1286,6 +1286,7 @@ int truncate_bdev_range(struct block_device *bdev, fmode_t mode, loff_t lstart,
 void invalidate_bdev(struct block_device *bdev);
 int sync_blockdev(struct block_device *bdev);
 int sync_blockdev_nowait(struct block_device *bdev);
+void sync_bdevs(bool wait);
 #else
 static inline void invalidate_bdev(struct block_device *bdev)
 {
@@ -1298,6 +1299,9 @@ static inline int sync_blockdev_nowait(struct block_device *bdev)
 {
 	return 0;
 }
+static inline void sync_bdevs(bool wait)
+{
+}
 #endif
 int fsync_bdev(struct block_device *bdev);
 
-- 
2.30.2