VFS: delay the dentry name generation on sockets and pipes

1) Introduces a new method in 'struct dentry_operations'.  This method
   called d_dname() might be called from d_path() to build a pathname for
   special filesystems.  It is called without locks.

   Future patches (if we succeed in having one common dentry for all
   pipes/sockets) may need to change prototype of this method, but we now
   use : char *d_dname(struct dentry *dentry, char *buffer, int buflen);

2) Adds a dynamic_dname() helper function that eases d_dname() implementations

3) Defines d_dname method for sockets : No more sprintf() at socket
   creation.  This is delayed up to the moment someone does an access to
   /proc/pid/fd/...

4) Defines d_dname method for pipes : No more sprintf() at pipe
   creation.  This is delayed up to the moment someone does an access to
   /proc/pid/fd/...

A benchmark consisting of 1.000.000 calls to pipe()/close()/close() gives a
*nice* speedup on my Pentium(M) 1.6 Ghz :

3.090 s instead of 3.450 s

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Acked-by: Christoph Hellwig <hch@infradead.org>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

diff --git a/Documentation/filesystems/Locking b/Documentation/filesystems/Locking
index 28bfea75bcf26151ac5594d131cbbc71d0dc93d0..59c14159cc47ddce04a19ae48d559fabedcf6902 100644 (file)
--- a/Documentation/filesystems/Locking
+++ b/Documentation/filesystems/Locking
@@ -15,6 +15,7 @@ prototypes:
        int (*d_delete)(struct dentry *);
        void (*d_release)(struct dentry *);
        void (*d_iput)(struct dentry *, struct inode *);
+       char *(*d_dname)((struct dentry *dentry, char *buffer, int buflen);
 
 locking rules:
        none have BKL
@@ -25,6 +26,7 @@ d_compare:    no              yes             no              no
 d_delete:      yes             no              yes             no
 d_release:     no              no              no              yes
 d_iput:                no              no              no              yes
+d_dname:       no              no              no              no
 
 --------------------------- inode_operations --------------------------- 
 prototypes:

diff --git a/Documentation/filesystems/vfs.txt b/Documentation/filesystems/vfs.txt
index ea271f2d395400b82cd8023db74fed76d878ce06..a47cc819f37bb7f5535afbf2bf720766ae51dfdd 100644 (file)
--- a/Documentation/filesystems/vfs.txt
+++ b/Documentation/filesystems/vfs.txt
@@ -827,7 +827,7 @@ This describes how a filesystem can overload the standard dentry
 operations. Dentries and the dcache are the domain of the VFS and the
 individual filesystem implementations. Device drivers have no business
 here. These methods may be set to NULL, as they are either optional or
-the VFS uses a default. As of kernel 2.6.13, the following members are
+the VFS uses a default. As of kernel 2.6.22, the following members are
 defined:
 
 struct dentry_operations {
@@ -837,6 +837,7 @@ struct dentry_operations {
        int (*d_delete)(struct dentry *);
        void (*d_release)(struct dentry *);
        void (*d_iput)(struct dentry *, struct inode *);
+       char *(*d_dname)(struct dentry *, char *, int);
 };
 
   d_revalidate: called when the VFS needs to revalidate a dentry. This
@@ -859,6 +860,26 @@ struct dentry_operations {
        VFS calls iput(). If you define this method, you must call
        iput() yourself
 
+  d_dname: called when the pathname of a dentry should be generated.
+       Usefull for some pseudo filesystems (sockfs, pipefs, ...) to delay
+       pathname generation. (Instead of doing it when dentry is created,
+       its done only when the path is needed.). Real filesystems probably
+       dont want to use it, because their dentries are present in global
+       dcache hash, so their hash should be an invariant. As no lock is
+       held, d_dname() should not try to modify the dentry itself, unless
+       appropriate SMP safety is used. CAUTION : d_path() logic is quite
+       tricky. The correct way to return for example "Hello" is to put it
+       at the end of the buffer, and returns a pointer to the first char.
+       dynamic_dname() helper function is provided to take care of this.
+
+Example :
+
+static char *pipefs_dname(struct dentry *dent, char *buffer, int buflen)
+{
+       return dynamic_dname(dentry, buffer, buflen, "pipe:[%lu]",
+                               dentry->d_inode->i_ino);
+}
+
 Each dentry has a pointer to its parent dentry, as well as a hash list
 of child dentries. Child dentries are basically like files in a
 directory.

diff --git a/fs/dcache.c b/fs/dcache.c
index 268da2e2bc09f8f4694dafb5a55675dd22b3c8ee..2135ab8bb103d1fabef4ef2f90b7bcafc50dbabd 100644 (file)
--- a/fs/dcache.c
+++ b/fs/dcache.c
@@ -1853,6 +1853,16 @@ char * d_path(struct dentry *dentry, struct vfsmount *vfsmnt,
        struct vfsmount *rootmnt;
        struct dentry *root;
 
+       /*
+        * We have various synthetic filesystems that never get mounted.  On
+        * these filesystems dentries are never used for lookup purposes, and
+        * thus don't need to be hashed.  They also don't need a name until a
+        * user wants to identify the object in /proc/pid/fd/.  The little hack
+        * below allows us to generate a name for these objects on demand:
+        */
+       if (dentry->d_op && dentry->d_op->d_dname)
+               return dentry->d_op->d_dname(dentry, buf, buflen);
+
        read_lock(&current->fs->lock);
        rootmnt = mntget(current->fs->rootmnt);
        root = dget(current->fs->root);
@@ -1865,6 +1875,27 @@ char * d_path(struct dentry *dentry, struct vfsmount *vfsmnt,
        return res;
 }
 
+/*
+ * Helper function for dentry_operations.d_dname() members
+ */
+char *dynamic_dname(struct dentry *dentry, char *buffer, int buflen,
+                       const char *fmt, ...)
+{
+       va_list args;
+       char temp[64];
+       int sz;
+
+       va_start(args, fmt);
+       sz = vsnprintf(temp, sizeof(temp), fmt, args) + 1;
+       va_end(args);
+
+       if (sz > sizeof(temp) || sz > buflen)
+               return ERR_PTR(-ENAMETOOLONG);
+
+       buffer += buflen - sz;
+       return memcpy(buffer, temp, sz);
+}
+
 /*
  * NOTE! The user-level library version returns a
  * character pointer. The kernel system call just

diff --git a/fs/pipe.c b/fs/pipe.c
index ebafde7d6abab948f9fa3a23e71f39360c757c71..3a89592bdf577930b4c1698c4811261aa467cbf2 100644 (file)
--- a/fs/pipe.c
+++ b/fs/pipe.c
@@ -841,8 +841,18 @@ static int pipefs_delete_dentry(struct dentry *dentry)
        return 0;
 }
 
+/*
+ * pipefs_dname() is called from d_path().
+ */
+static char *pipefs_dname(struct dentry *dentry, char *buffer, int buflen)
+{
+       return dynamic_dname(dentry, buffer, buflen, "pipe:[%lu]",
+                               dentry->d_inode->i_ino);
+}
+
 static struct dentry_operations pipefs_dentry_operations = {
        .d_delete       = pipefs_delete_dentry,
+       .d_dname        = pipefs_dname,
 };
 
 static struct inode * get_pipe_inode(void)
@@ -888,8 +898,7 @@ struct file *create_write_pipe(void)
        struct inode *inode;
        struct file *f;
        struct dentry *dentry;
-       char name[32];
-       struct qstr this;
+       struct qstr name = { .name = "" };
 
        f = get_empty_filp();
        if (!f)
@@ -899,11 +908,8 @@ struct file *create_write_pipe(void)
        if (!inode)
                goto err_file;
 
-       this.len = sprintf(name, "[%lu]", inode->i_ino);
-       this.name = name;
-       this.hash = 0;
        err = -ENOMEM;
-       dentry = d_alloc(pipe_mnt->mnt_sb->s_root, &this);
+       dentry = d_alloc(pipe_mnt->mnt_sb->s_root, &name);
        if (!dentry)
                goto err_inode;
 

diff --git a/include/linux/dcache.h b/include/linux/dcache.h
index 63f64a9a5bf7b207b674be433d1c3b218ff39813..aab53df4fafa338e46b96919d61e77ed2b20ff78 100644 (file)
--- a/include/linux/dcache.h
+++ b/include/linux/dcache.h
@@ -133,6 +133,7 @@ struct dentry_operations {
        int (*d_delete)(struct dentry *);
        void (*d_release)(struct dentry *);
        void (*d_iput)(struct dentry *, struct inode *);
+       char *(*d_dname)(struct dentry *, char *, int);
 };
 
 /* the dentry parameter passed to d_hash and d_compare is the parent
@@ -293,6 +294,11 @@ extern struct dentry * d_hash_and_lookup(struct dentry *, struct qstr *);
 /* validate "insecure" dentry pointer */
 extern int d_validate(struct dentry *, struct dentry *);
 
+/*
+ * helper function for dentry_operations.d_dname() members
+ */
+extern char *dynamic_dname(struct dentry *, char *, int, const char *, ...);
+
 extern char * d_path(struct dentry *, struct vfsmount *, char *, int);
   
 /* Allocation counts.. */

diff --git a/net/socket.c b/net/socket.c
index 759825b7ca263b4073d39338debc7a4a04a4a6a1..98a8f67abbfcfe84f2cb026737c710c39647d8cc 100644 (file)
--- a/net/socket.c
+++ b/net/socket.c
@@ -313,8 +313,19 @@ static int sockfs_delete_dentry(struct dentry *dentry)
        dentry->d_flags |= DCACHE_UNHASHED;
        return 0;
 }
+
+/*
+ * sockfs_dname() is called from d_path().
+ */
+static char *sockfs_dname(struct dentry *dentry, char *buffer, int buflen)
+{
+       return dynamic_dname(dentry, buffer, buflen, "socket:[%lu]",
+                               dentry->d_inode->i_ino);
+}
+
 static struct dentry_operations sockfs_dentry_operations = {
        .d_delete = sockfs_delete_dentry,
+       .d_dname  = sockfs_dname,
 };
 
 /*
@@ -354,14 +365,9 @@ static int sock_alloc_fd(struct file **filep)
 
 static int sock_attach_fd(struct socket *sock, struct file *file)
 {
-       struct qstr this;
-       char name[32];
-
-       this.len = sprintf(name, "[%lu]", SOCK_INODE(sock)->i_ino);
-       this.name = name;
-       this.hash = 0;
+       struct qstr name = { .name = "" };
 
-       file->f_path.dentry = d_alloc(sock_mnt->mnt_sb->s_root, &this);
+       file->f_path.dentry = d_alloc(sock_mnt->mnt_sb->s_root, &name);
        if (unlikely(!file->f_path.dentry))
                return -ENOMEM;