vfs: optimize inode cache access patterns

The inode structure layout is largely random, and some of the vfs paths
really do care.  The path lookup in particular is already quite D$
intensive, and profiles show that accessing the 'inode->i_op->xyz'
fields is quite costly.

We already optimized the dcache to not unnecessarily load the d_op
structure for members that are often NULL using the DCACHE_OP_xyz bits
in dentry->d_flags, and this does something very similar for the inode
ops that are used during pathname lookup.

It also re-orders the fields so that the fields accessed by 'stat' are
together at the beginning of the inode structure, and roughly in the
order accessed.

The effect of this seems to be in the 1-2% range for an empty kernel
"make -j" run (which is fairly kernel-intensive, mostly in filename
lookup), so it's visible.  The numbers are fairly noisy, though, and
likely depend a lot on exact microarchitecture.  So there's more tuning
to be done.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

1 files changed, 37 insertions, 22 deletions

diff --git a/include/linux/fs.h b/include/linux/fs.h
index 786b3b1..178cdb4 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -738,22 +738,54 @@ static inline int mapping_writably_mapped(struct address_space *mapping)
 struct posix_acl;
 #define ACL_NOT_CACHED ((void *)(-1))
 
+#define IOP_FASTPERM	0x0001
+#define IOP_LOOKUP	0x0002
+#define IOP_NOFOLLOW	0x0004
+
+/*
+ * Keep mostly read-only and often accessed (especially for
+ * the RCU path lookup and 'stat' data) fields at the beginning
+ * of the 'struct inode'
+ */
 struct inode {
-	/* RCU path lookup touches following: */
 	umode_t			i_mode;
+	unsigned short		i_opflags;
 	uid_t			i_uid;
 	gid_t			i_gid;
+	unsigned int		i_flags;
+
+#ifdef CONFIG_FS_POSIX_ACL
+	struct posix_acl	*i_acl;
+	struct posix_acl	*i_default_acl;
+#endif
+
 	const struct inode_operations	*i_op;
 	struct super_block	*i_sb;
+	struct address_space	*i_mapping;
 
-	spinlock_t		i_lock;	/* i_blocks, i_bytes, maybe i_size */
-	unsigned int		i_flags;
-	unsigned long		i_state;
 #ifdef CONFIG_SECURITY
 	void			*i_security;
 #endif
-	struct mutex		i_mutex;
 
+	/* Stat data, not accessed from path walking */
+	unsigned long		i_ino;
+	unsigned int		i_nlink;
+	dev_t			i_rdev;
+	loff_t			i_size;
+	struct timespec		i_atime;
+	struct timespec		i_mtime;
+	struct timespec		i_ctime;
+	unsigned int		i_blkbits;
+	blkcnt_t		i_blocks;
+
+#ifdef __NEED_I_SIZE_ORDERED
+	seqcount_t		i_size_seqcount;
+#endif
+
+	/* Misc */
+	unsigned long		i_state;
+	spinlock_t		i_lock;	/* i_blocks, i_bytes, maybe i_size */
+	struct mutex		i_mutex;
 
 	unsigned long		dirtied_when;	/* jiffies of first dirtying */
 
@@ -765,25 +797,12 @@ struct inode {
 		struct list_head	i_dentry;
 		struct rcu_head		i_rcu;
 	};
-	unsigned long		i_ino;
 	atomic_t		i_count;
-	unsigned int		i_nlink;
-	dev_t			i_rdev;
-	unsigned int		i_blkbits;
 	u64			i_version;
-	loff_t			i_size;
-#ifdef __NEED_I_SIZE_ORDERED
-	seqcount_t		i_size_seqcount;
-#endif
-	struct timespec		i_atime;
-	struct timespec		i_mtime;
-	struct timespec		i_ctime;
-	blkcnt_t		i_blocks;
 	unsigned short          i_bytes;
 	atomic_t		i_dio_count;
 	const struct file_operations	*i_fop;	/* former ->i_op->default_file_ops */
 	struct file_lock	*i_flock;
-	struct address_space	*i_mapping;
 	struct address_space	i_data;
 #ifdef CONFIG_QUOTA
 	struct dquot		*i_dquot[MAXQUOTAS];
@@ -806,10 +825,6 @@ struct inode {
 	atomic_t		i_readcount; /* struct files open RO */
 #endif
 	atomic_t		i_writecount;
-#ifdef CONFIG_FS_POSIX_ACL
-	struct posix_acl	*i_acl;
-	struct posix_acl	*i_default_acl;
-#endif
 	void			*i_private; /* fs or device private pointer */
 };