[PATCH v4] nilfs2: implement calculation of free inodes count

[PATCH v4] nilfs2: implement calculation of free inodes count

[Vyacheslav Dubeyko]

Hi Ryusuke,

This is fourth version of the patch. I reworked representation of
mi_desc_blocks_count because I think that using this field is good
solution for calculation of max entries count.

By the way, I see that it is used atomic_t type for representation
of inodes_count and blocks_count fields in nilfs_root structure.
Maybe it makes sense to rework representation of these fields too.
What do you think?

v3->v4
 * The desc_blocks value is represented by unsigned long type.
 * The nilfs_palloc_count_desc_blocks() method is simplified.
 * The mi_desc_blocks_count field of nilfs_mdt_info structure
   is represented as atomic64_t for the case of CONFIG_64BIT and
   as atomic_t for other cases.

v2->v3
 * Trivial BUG_ON checks were removed.
 * Whole calculation algorithm was moved into
   nilfs_palloc_count_max_entries() method.
 * Improve error processing in the code.
 * The nilfs_palloc_mdt_file_can_grow() method was simplified.
 * The nilfs_ifile_count_free_inodes() method was simplified.

v1->v2
 * Change __statfs_word on u64 type.
 * Rename nilfs_count_free_inodes() into nilfs_ifile_count_free_inodes()
   method.
 * Introduce auxiliary functions: nilfs_palloc_count_max_entries(),
   nilfs_palloc_count_desc_blocks(), nilfs_palloc_mdt_file_can_grow().
 * Rework processing of returned error from nilfs_ifile_count_free_inodes()
   in nilfs_statfs().

With the best regards,
Vyacheslav Dubeyko.
---
From: Vyacheslav Dubeyko <slava@dubeyko.com>
Subject: [PATCH v4] nilfs2: implement calculation of free inodes count

Currently, NILFS2 returns 0 as free inodes count (f_ffree) and
current used inodes count as total file nodes in file system
(f_files):

df -i
Filesystem      Inodes  IUsed   IFree IUse% Mounted on
/dev/loop0           2      2       0  100% /mnt/nilfs2

This patch implements real calculation of free inodes count.
First of all, it is calculated total file nodes in file system
as (desc_blocks_count * groups_per_desc_block * entries_per_group).
Then, it is calculated free inodes count as difference the total
file nodes and used inodes count. As a result, we have such output
for NILFS2:

df -i
Filesystem       Inodes   IUsed    IFree IUse% Mounted on
/dev/loop0      4194304 2114701  2079603   51% /mnt/nilfs2

Reported-by: Clemens Eisserer <linuxhippy@gmail.com>
Signed-off-by: Vyacheslav Dubeyko <slava@dubeyko.com>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Tested-by: Vyacheslav Dubeyko <slava@dubeyko.com>
---
 fs/nilfs2/alloc.c |   78 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 fs/nilfs2/alloc.h |    2 ++
 fs/nilfs2/ifile.c |   22 +++++++++++++++
 fs/nilfs2/ifile.h |    2 +-
 fs/nilfs2/mdt.h   |   26 ++++++++++++++++++
 fs/nilfs2/super.c |   25 +++++++++++++++--
 6 files changed, 152 insertions(+), 3 deletions(-)

diff --git a/fs/nilfs2/alloc.c b/fs/nilfs2/alloc.c
index eed4d7b..e5ea7b3 100644
--- a/fs/nilfs2/alloc.c
+++ b/fs/nilfs2/alloc.c
@@ -80,6 +80,13 @@ int nilfs_palloc_init_blockgroup(struct inode *inode, unsigned entry_size)
 		mi->mi_blocks_per_group + 1;
 		/* Number of blocks per descriptor including the
 		   descriptor block */
+	nilfs_mdt_set_desc_blocks(mi, 1);
+		/*
+		 * The field mi_desc_blocks_count is used for
+		 * storing knowledge about current count of
+		 * desciptor blocks. Initially, it is initialized
+		 * by one.
+		 */
 	return 0;
 }
 
@@ -398,6 +405,77 @@ nilfs_palloc_rest_groups_in_desc_block(const struct inode *inode,
 }
 
 /**
+ * nilfs_palloc_count_desc_blocks - count descriptor blocks number
+ * @inode: inode of metadata file using this allocator
+ * @desc_blocks: descriptor blocks number [out]
+ */
+static int nilfs_palloc_count_desc_blocks(struct inode *inode,
+					  unsigned long *desc_blocks)
+{
+	unsigned long blknum;
+	int ret;
+
+	ret = nilfs_bmap_last_key(NILFS_I(inode)->i_bmap, &blknum);
+	if (likely(!ret))
+		*desc_blocks = DIV_ROUND_UP(
+			blknum, NILFS_MDT(inode)->mi_blocks_per_desc_block);
+	return ret;
+}
+
+/**
+ * nilfs_palloc_mdt_file_can_grow - check potential opportunity for
+ *					MDT file growing
+ * @inode: inode of metadata file using this allocator
+ * @desc_blocks: known current descriptor blocks count
+ */
+static inline bool nilfs_palloc_mdt_file_can_grow(struct inode *inode,
+						    unsigned long desc_blocks)
+{
+	return (nilfs_palloc_groups_per_desc_block(inode) * desc_blocks) <
+			nilfs_palloc_groups_count(inode);
+}
+
+/**
+ * nilfs_palloc_count_max_entries - count max number of entries that can be
+ *					described by descriptor blocks count
+ * @inode: inode of metadata file using this allocator
+ * @nused: current number of used entries
+ * @nmaxp: max number of entries [out]
+ */
+int nilfs_palloc_count_max_entries(struct inode *inode, u64 nused, u64 *nmaxp)
+{
+	unsigned long desc_blocks;
+	u64 entries_per_desc_block, nmax;
+	int err;
+
+	desc_blocks = nilfs_mdt_read_desc_blocks(NILFS_MDT(inode));
+	entries_per_desc_block = (u64)nilfs_palloc_entries_per_group(inode) *
+				nilfs_palloc_groups_per_desc_block(inode);
+
+	nmax = entries_per_desc_block * desc_blocks;
+	if (nused >= nmax) {
+		err = nilfs_palloc_count_desc_blocks(inode, &desc_blocks);
+		if (unlikely(err))
+			return err;
+
+		nmax = entries_per_desc_block * desc_blocks;
+		if (nused == nmax &&
+				nilfs_palloc_mdt_file_can_grow(inode,
+								desc_blocks)) {
+			nmax += entries_per_desc_block;
+			++desc_blocks;
+		}
+
+		nilfs_mdt_set_desc_blocks(NILFS_MDT(inode), desc_blocks);
+
+		if (nused > nmax)
+			return -ERANGE;
+	}
+	*nmaxp = nmax;
+	return 0;
+}
+
+/**
  * nilfs_palloc_prepare_alloc_entry - prepare to allocate a persistent object
  * @inode: inode of metadata file using this allocator
  * @req: nilfs_palloc_req structure exchanged for the allocation
diff --git a/fs/nilfs2/alloc.h b/fs/nilfs2/alloc.h
index fb72381..4bd6451 100644
--- a/fs/nilfs2/alloc.h
+++ b/fs/nilfs2/alloc.h
@@ -48,6 +48,8 @@ int nilfs_palloc_get_entry_block(struct inode *, __u64, int,
 void *nilfs_palloc_block_get_entry(const struct inode *, __u64,
 				   const struct buffer_head *, void *);
 
+int nilfs_palloc_count_max_entries(struct inode *, u64, u64 *);
+
 /**
  * nilfs_palloc_req - persistent allocator request and reply
  * @pr_entry_nr: entry number (vblocknr or inode number)
diff --git a/fs/nilfs2/ifile.c b/fs/nilfs2/ifile.c
index d8e65bd..d788a59 100644
--- a/fs/nilfs2/ifile.c
+++ b/fs/nilfs2/ifile.c
@@ -160,6 +160,28 @@ int nilfs_ifile_get_inode_block(struct inode *ifile, ino_t ino,
 }
 
 /**
+ * nilfs_ifile_count_free_inodes - calculate free inodes count
+ * @ifile: ifile inode
+ * @nmaxinodes: current maximum of available inodes count [out]
+ * @nfreeinodes: free inodes count [out]
+ */
+int nilfs_ifile_count_free_inodes(struct inode *ifile,
+				    u64 *nmaxinodes, u64 *nfreeinodes)
+{
+	u64 nused;
+	int err;
+
+	*nmaxinodes = 0;
+	*nfreeinodes = 0;
+
+	nused = atomic_read(&NILFS_I(ifile)->i_root->inodes_count);
+	err = nilfs_palloc_count_max_entries(ifile, nused, nmaxinodes);
+	if (likely(!err))
+		*nfreeinodes = *nmaxinodes - nused;
+	return err;
+}
+
+/**
  * nilfs_ifile_read - read or get ifile inode
  * @sb: super block instance
  * @root: root object
diff --git a/fs/nilfs2/ifile.h b/fs/nilfs2/ifile.h
index 59b6f2b..e9ceb27 100644
--- a/fs/nilfs2/ifile.h
+++ b/fs/nilfs2/ifile.h
@@ -48,7 +48,7 @@ static inline void nilfs_ifile_unmap_inode(struct inode *ifile, ino_t ino,
 int nilfs_ifile_create_inode(struct inode *, ino_t *, struct buffer_head **);
 int nilfs_ifile_delete_inode(struct inode *, ino_t);
 int nilfs_ifile_get_inode_block(struct inode *, ino_t, struct buffer_head **);
-
+int nilfs_ifile_count_free_inodes(struct inode *, u64 *, u64 *);
 int nilfs_ifile_read(struct super_block *sb, struct nilfs_root *root,
 		     size_t inode_size, struct nilfs_inode *raw_inode,
 		     struct inode **inodep);
diff --git a/fs/nilfs2/mdt.h b/fs/nilfs2/mdt.h
index ab172e8..bcd11bf 100644
--- a/fs/nilfs2/mdt.h
+++ b/fs/nilfs2/mdt.h
@@ -53,6 +53,7 @@ struct nilfs_shadow_map {
  * @mi_shadow: shadow of bmap and page caches
  * @mi_blocks_per_group: number of blocks in a group
  * @mi_blocks_per_desc_block: number of blocks per descriptor block
+ * @mi_desc_blocks_count: number of descriptor blocks
  */
 struct nilfs_mdt_info {
 	struct rw_semaphore	mi_sem;
@@ -64,8 +65,33 @@ struct nilfs_mdt_info {
 	struct nilfs_shadow_map *mi_shadow;
 	unsigned long		mi_blocks_per_group;
 	unsigned long		mi_blocks_per_desc_block;
+#ifdef CONFIG_64BIT
+	atomic64_t		mi_desc_blocks_count;
+#else
+	atomic_t		mi_desc_blocks_count;
+#endif /* CONFIG_64BIT */
 };
 
+static inline
+void nilfs_mdt_set_desc_blocks(struct nilfs_mdt_info *mi, unsigned long value)
+{
+#ifdef CONFIG_64BIT
+	atomic64_set(&mi->mi_desc_blocks_count, value);
+#else
+	atomic_set(&mi->mi_desc_blocks_count, value);
+#endif
+}
+
+static inline
+unsigned long nilfs_mdt_read_desc_blocks(struct nilfs_mdt_info *mi)
+{
+#ifdef CONFIG_64BIT
+	return atomic64_read(&mi->mi_desc_blocks_count);
+#else
+	return atomic_read(&mi->mi_desc_blocks_count);
+#endif
+}
+
 static inline struct nilfs_mdt_info *NILFS_MDT(const struct inode *inode)
 {
 	return inode->i_private;
diff --git a/fs/nilfs2/super.c b/fs/nilfs2/super.c
index c7d1f9f..acd110f 100644
--- a/fs/nilfs2/super.c
+++ b/fs/nilfs2/super.c
@@ -609,6 +609,7 @@ static int nilfs_statfs(struct dentry *dentry, struct kstatfs *buf)
 	unsigned long overhead;
 	unsigned long nrsvblocks;
 	sector_t nfreeblocks;
+	u64 nmaxinodes, nfreeinodes;
 	int err;
 
 	/*
@@ -633,14 +634,34 @@ static int nilfs_statfs(struct dentry *dentry, struct kstatfs *buf)
 	if (unlikely(err))
 		return err;
 
+	err = nilfs_ifile_count_free_inodes(root->ifile,
+						&nmaxinodes, &nfreeinodes);
+	if (unlikely(err)) {
+		printk(KERN_WARNING
+			"NILFS warning: fail to count free inodes: err %d.\n",
+			err);
+		if (err == -ERANGE) {
+			/*
+			 * If nilfs_palloc_count_max_entries() returns
+			 * -ERANGE error code then we simply treat
+			 * curent inodes count as maximum possible and
+			 * zero as free inodes value.
+			 */
+			nmaxinodes = atomic_read(&root->inodes_count);
+			nfreeinodes = 0;
+			err = 0;
+		} else
+			return err;
+	}
+
 	buf->f_type = NILFS_SUPER_MAGIC;
 	buf->f_bsize = sb->s_blocksize;
 	buf->f_blocks = blocks - overhead;
 	buf->f_bfree = nfreeblocks;
 	buf->f_bavail = (buf->f_bfree >= nrsvblocks) ?
 		(buf->f_bfree - nrsvblocks) : 0;
-	buf->f_files = atomic_read(&root->inodes_count);
-	buf->f_ffree = 0; /* nilfs_count_free_inodes(sb); */
+	buf->f_files = nmaxinodes;
+	buf->f_ffree = nfreeinodes;
 	buf->f_namelen = NILFS_NAME_LEN;
 	buf->f_fsid.val[0] = (u32)id;
 	buf->f_fsid.val[1] = (u32)(id >> 32);
-- 
1.7.9.5

Re: [PATCH v4] nilfs2: implement calculation of free inodes count

[Ryusuke Konishi]

On Wed, 22 May 2013 10:23:47 +0400, Vyacheslav Dubeyko wrote:
> Hi Ryusuke,
> 
> This is fourth version of the patch. I reworked representation of
> mi_desc_blocks_count because I think that using this field is good
> solution for calculation of max entries count.

Nack.

Please stop switching atomic_t and atomic64_t in nilfs with heavy use
of ifdef macros and inline functions.  It is seriously hurting the
readability and simplicity of source code.

Consider using "atomic_long_t", and please try to keep the patch
simple.

In my opinion, mi_desc_blocks_count is no longer needed because
nilfs_bmap_last_key is not so heavy.  But, I don't deny adding this
counter.
 

> By the way, I see that it is used atomic_t type for representation
> of inodes_count and blocks_count fields in nilfs_root structure.
> Maybe it makes sense to rework representation of these fields too.
> What do you think?

Yes, looks like they should be replaced with atomic_long_t type
counters (with a separate patch, of course).


Regards,
Ryusuke Konishi


> v3->v4
>  * The desc_blocks value is represented by unsigned long type.
>  * The nilfs_palloc_count_desc_blocks() method is simplified.
>  * The mi_desc_blocks_count field of nilfs_mdt_info structure
>    is represented as atomic64_t for the case of CONFIG_64BIT and
>    as atomic_t for other cases.
> 
> v2->v3
>  * Trivial BUG_ON checks were removed.
>  * Whole calculation algorithm was moved into
>    nilfs_palloc_count_max_entries() method.
>  * Improve error processing in the code.
>  * The nilfs_palloc_mdt_file_can_grow() method was simplified.
>  * The nilfs_ifile_count_free_inodes() method was simplified.
> 
> v1->v2
>  * Change __statfs_word on u64 type.
>  * Rename nilfs_count_free_inodes() into nilfs_ifile_count_free_inodes()
>    method.
>  * Introduce auxiliary functions: nilfs_palloc_count_max_entries(),
>    nilfs_palloc_count_desc_blocks(), nilfs_palloc_mdt_file_can_grow().
>  * Rework processing of returned error from nilfs_ifile_count_free_inodes()
>    in nilfs_statfs().
> 
> With the best regards,
> Vyacheslav Dubeyko.
> ---
> From: Vyacheslav Dubeyko <slava-yeENwD64cLxBDgjK7y7TUQ@public.gmane.org>
> Subject: [PATCH v4] nilfs2: implement calculation of free inodes count
> 
> Currently, NILFS2 returns 0 as free inodes count (f_ffree) and
> current used inodes count as total file nodes in file system
> (f_files):
> 
> df -i
> Filesystem      Inodes  IUsed   IFree IUse% Mounted on
> /dev/loop0           2      2       0  100% /mnt/nilfs2
> 
> This patch implements real calculation of free inodes count.
> First of all, it is calculated total file nodes in file system
> as (desc_blocks_count * groups_per_desc_block * entries_per_group).
> Then, it is calculated free inodes count as difference the total
> file nodes and used inodes count. As a result, we have such output
> for NILFS2:
> 
> df -i
> Filesystem       Inodes   IUsed    IFree IUse% Mounted on
> /dev/loop0      4194304 2114701  2079603   51% /mnt/nilfs2
> 
> Reported-by: Clemens Eisserer <linuxhippy-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
> Signed-off-by: Vyacheslav Dubeyko <slava-yeENwD64cLxBDgjK7y7TUQ@public.gmane.org>
> Signed-off-by: Ryusuke Konishi <konishi.ryusuke-Zyj7fXuS5i5L9jVzuh4AOg@public.gmane.org>
> Tested-by: Vyacheslav Dubeyko <slava-yeENwD64cLxBDgjK7y7TUQ@public.gmane.org>
> ---
>  fs/nilfs2/alloc.c |   78 +++++++++++++++++++++++++++++++++++++++++++++++++++++
>  fs/nilfs2/alloc.h |    2 ++
>  fs/nilfs2/ifile.c |   22 +++++++++++++++
>  fs/nilfs2/ifile.h |    2 +-
>  fs/nilfs2/mdt.h   |   26 ++++++++++++++++++
>  fs/nilfs2/super.c |   25 +++++++++++++++--
>  6 files changed, 152 insertions(+), 3 deletions(-)
> 
> diff --git a/fs/nilfs2/alloc.c b/fs/nilfs2/alloc.c
> index eed4d7b..e5ea7b3 100644
> --- a/fs/nilfs2/alloc.c
> +++ b/fs/nilfs2/alloc.c
> @@ -80,6 +80,13 @@ int nilfs_palloc_init_blockgroup(struct inode *inode, unsigned entry_size)
>  		mi->mi_blocks_per_group + 1;
>  		/* Number of blocks per descriptor including the
>  		   descriptor block */
> +	nilfs_mdt_set_desc_blocks(mi, 1);
> +		/*
> +		 * The field mi_desc_blocks_count is used for
> +		 * storing knowledge about current count of
> +		 * desciptor blocks. Initially, it is initialized
> +		 * by one.
> +		 */
>  	return 0;
>  }
>  
> @@ -398,6 +405,77 @@ nilfs_palloc_rest_groups_in_desc_block(const struct inode *inode,
>  }
>  
>  /**
> + * nilfs_palloc_count_desc_blocks - count descriptor blocks number
> + * @inode: inode of metadata file using this allocator
> + * @desc_blocks: descriptor blocks number [out]
> + */
> +static int nilfs_palloc_count_desc_blocks(struct inode *inode,
> +					  unsigned long *desc_blocks)
> +{
> +	unsigned long blknum;
> +	int ret;
> +
> +	ret = nilfs_bmap_last_key(NILFS_I(inode)->i_bmap, &blknum);
> +	if (likely(!ret))
> +		*desc_blocks = DIV_ROUND_UP(
> +			blknum, NILFS_MDT(inode)->mi_blocks_per_desc_block);
> +	return ret;
> +}
> +
> +/**
> + * nilfs_palloc_mdt_file_can_grow - check potential opportunity for
> + *					MDT file growing
> + * @inode: inode of metadata file using this allocator
> + * @desc_blocks: known current descriptor blocks count
> + */
> +static inline bool nilfs_palloc_mdt_file_can_grow(struct inode *inode,
> +						    unsigned long desc_blocks)
> +{
> +	return (nilfs_palloc_groups_per_desc_block(inode) * desc_blocks) <
> +			nilfs_palloc_groups_count(inode);
> +}
> +
> +/**
> + * nilfs_palloc_count_max_entries - count max number of entries that can be
> + *					described by descriptor blocks count
> + * @inode: inode of metadata file using this allocator
> + * @nused: current number of used entries
> + * @nmaxp: max number of entries [out]
> + */
> +int nilfs_palloc_count_max_entries(struct inode *inode, u64 nused, u64 *nmaxp)
> +{
> +	unsigned long desc_blocks;
> +	u64 entries_per_desc_block, nmax;
> +	int err;
> +
> +	desc_blocks = nilfs_mdt_read_desc_blocks(NILFS_MDT(inode));
> +	entries_per_desc_block = (u64)nilfs_palloc_entries_per_group(inode) *
> +				nilfs_palloc_groups_per_desc_block(inode);
> +
> +	nmax = entries_per_desc_block * desc_blocks;
<snip>
--
To unsubscribe from this list: send the line "unsubscribe linux-nilfs" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html