fs: cache optimise dentry and inode for rcu-walk

Put dentry and inode fields into top of data structure.  This allows RCU path
traversal to perform an RCU dentry lookup in a path walk by touching only the
first 56 bytes of the dentry.

We also fit in 8 bytes of inline name in the first 64 bytes, so for short
names, only 64 bytes needs to be touched to perform the lookup. We should
get rid of the hash->prev pointer from the first 64 bytes, and fit 16 bytes
of name in there, which will take care of 81% rather than 32% of the kernel
tree.

inode is also rearranged so that RCU lookup will only touch a single cacheline
in the inode, plus one in the i_ops structure.

This is important for directory component lookups in RCU path walking. In the
kernel source, directory names average is around 6 chars, so this works.

When we reach the last element of the lookup, we need to lock it and take its
refcount which requires another cacheline access.

Align dentry and inode operations structs, so members will be at predictable
offsets and we can group common operations into head of structure.

Signed-off-by: Nick Piggin <npiggin@kernel.dk>

1 files changed, 19 insertions, 17 deletions

diff --git a/include/linux/dcache.h b/include/linux/dcache.h
index f4b40a751f09..b1aeda077258 100644
--- a/include/linux/dcache.h
+++ b/include/linux/dcache.h
@@ -82,25 +82,33 @@ full_name_hash(const unsigned char *name, unsigned int len)
  * large memory footprint increase).
  */
 #ifdef CONFIG_64BIT
-#define DNAME_INLINE_LEN_MIN 32 /* 192 bytes */
+# define DNAME_INLINE_LEN 32 /* 192 bytes */
 #else
-#define DNAME_INLINE_LEN_MIN 40 /* 128 bytes */
+# ifdef CONFIG_SMP
+#  define DNAME_INLINE_LEN 36 /* 128 bytes */
+# else
+#  define DNAME_INLINE_LEN 40 /* 128 bytes */
+# endif
 #endif
 
 struct dentry {
-        unsigned int d_count;           /* protected by d_lock */
+        /* RCU lookup touched fields */
         unsigned int d_flags;           /* protected by d_lock */
-        spinlock_t d_lock;              /* per dentry lock */
         seqcount_t d_seq;               /* per dentry seqlock */
-        struct inode *d_inode;          /* Where the name belongs to - NULL is
-                                         * negative */
-        /*
-         * The next three fields are touched by __d_lookup.  Place them here
-         * so they all fit in a cache line.
-         */
         struct hlist_node d_hash;       /* lookup hash list */
         struct dentry *d_parent;        /* parent directory */
         struct qstr d_name;
+        struct inode *d_inode;          /* Where the name belongs to - NULL is
+                                         * negative */
+        unsigned char d_iname[DNAME_INLINE_LEN];        /* small names */
+
+        /* Ref lookup also touches following */
+        unsigned int d_count;           /* protected by d_lock */
+        spinlock_t d_lock;              /* per dentry lock */
+        const struct dentry_operations *d_op;
+        struct super_block *d_sb;       /* The root of the dentry tree */
+        unsigned long d_time;           /* used by d_revalidate */
+        void *d_fsdata;                 /* fs-specific data */
 
         struct list_head d_lru;         /* LRU list */
         /*
@@ -112,12 +120,6 @@ struct dentry {
         } d_u;
         struct list_head d_subdirs;     /* our children */
         struct list_head d_alias;       /* inode alias list */
-        unsigned long d_time;           /* used by d_revalidate */
-        const struct dentry_operations *d_op;
-        struct super_block *d_sb;       /* The root of the dentry tree */
-        void *d_fsdata;                 /* fs-specific data */
-
-        unsigned char d_iname[DNAME_INLINE_LEN_MIN];    /* small names */
 };
 
 /*
@@ -143,7 +145,7 @@ struct dentry_operations {
         void (*d_release)(struct dentry *);
         void (*d_iput)(struct dentry *, struct inode *);
         char *(*d_dname)(struct dentry *, char *, int);
-};
+} ____cacheline_aligned;
 
 /*
  * Locking rules for dentry_operations callbacks are to be found in