From c23fbb6bcb3eb9cdf39a103edadf57bde8ce309c Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 8 May 2007 00:26:18 -0700 Subject: [PATCH] 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 Acked-by: Christoph Hellwig Acked-by: Linus Torvalds Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- Documentation/filesystems/Locking | 2 ++ Documentation/filesystems/vfs.txt | 23 ++++++++++++++++++++++- fs/dcache.c | 31 +++++++++++++++++++++++++++++++ fs/pipe.c | 18 ++++++++++++------ include/linux/dcache.h | 6 ++++++ net/socket.c | 20 +++++++++++++------- 6 files changed, 86 insertions(+), 14 deletions(-) diff --git a/Documentation/filesystems/Locking b/Documentation/filesystems/Locking index 28bfea75bcf2..59c14159cc47 100644 --- 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 ea271f2d3954..a47cc819f37b 100644 --- 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 268da2e2bc09..2135ab8bb103 100644 --- 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(¤t->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 ebafde7d6aba..3a89592bdf57 100644 --- 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 63f64a9a5bf7..aab53df4fafa 100644 --- 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 759825b7ca26..98a8f67abbfc 100644 --- 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]; + struct qstr name = { .name = "" }; - this.len = sprintf(name, "[%lu]", SOCK_INODE(sock)->i_ino); - this.name = name; - this.hash = 0; - - 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;