[PATCH] vfs: *at functions: core

Here is a series of patches which introduce in total 13 new system calls
which take a file descriptor/filename pair instead of a single file
name.  These functions, openat etc, have been discussed on numerous
occasions.  They are needed to implement race-free filesystem traversal,
they are necessary to implement a virtual per-thread current working
directory (think multi-threaded backup software), etc.

We have in glibc today implementations of the interfaces which use the
/proc/self/fd magic.  But this code is rather expensive.  Here are some
results (similar to what Jim Meyering posted before).

The test creates a deep directory hierarchy on a tmpfs filesystem.  Then
rm -fr is used to remove all directories.  Without syscall support I get
this:

real    0m31.921s
user    0m0.688s
sys     0m31.234s

With syscall support the results are much better:

real    0m20.699s
user    0m0.536s
sys     0m20.149s

The interfaces are for obvious reasons currently not much used.  But they'll
be used.  coreutils (and Jeff's posixutils) are already using them.
Furthermore, code like ftw/fts in libc (maybe even glob) will also start using
them.  I expect a patch to make follow soon.  Every program which is walking
the filesystem tree will benefit.

Signed-off-by: Ulrich Drepper <drepper@redhat.com>
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Al Viro <viro@ftp.linux.org.uk>
Acked-by: Ingo Molnar <mingo@elte.hu>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

1 files changed, 68 insertions, 15 deletions

diff --git a/fs/open.c b/fs/open.c
index 8e20c1f3256..70e0230d8e7 100644
--- a/fs/open.c
+++ b/fs/open.c
@@ -20,6 +20,7 @@
 #include <linux/security.h>
 #include <linux/mount.h>
 #include <linux/vfs.h>
+#include <linux/fcntl.h>
 #include <asm/uaccess.h>
 #include <linux/fs.h>
 #include <linux/personality.h>
@@ -383,7 +384,7 @@ asmlinkage long sys_utime(char __user * filename, struct utimbuf __user * times)
 
 		error = get_user(newattrs.ia_atime.tv_sec, &times->actime);
 		newattrs.ia_atime.tv_nsec = 0;
-		if (!error) 
+		if (!error)
 			error = get_user(newattrs.ia_mtime.tv_sec, &times->modtime);
 		newattrs.ia_mtime.tv_nsec = 0;
 		if (error)
@@ -414,14 +415,14 @@ out:
  * must be owner or have write permission.
  * Else, update from *times, must be owner or super user.
  */
-long do_utimes(char __user * filename, struct timeval * times)
+long do_utimes(int dfd, char __user *filename, struct timeval *times)
 {
 	int error;
 	struct nameidata nd;
 	struct inode * inode;
 	struct iattr newattrs;
 
-	error = user_path_walk(filename, &nd);
+	error = __user_walk_fd(dfd, filename, LOOKUP_FOLLOW, &nd);
 
 	if (error)
 		goto out;
@@ -461,13 +462,18 @@ out:
 	return error;
 }
 
-asmlinkage long sys_utimes(char __user * filename, struct timeval __user * utimes)
+asmlinkage long sys_futimesat(int dfd, char __user *filename, struct timeval __user *utimes)
 {
 	struct timeval times[2];
 
 	if (utimes && copy_from_user(&times, utimes, sizeof(times)))
 		return -EFAULT;
-	return do_utimes(filename, utimes ? times : NULL);
+	return do_utimes(dfd, filename, utimes ? times : NULL);
+}
+
+asmlinkage long sys_utimes(char __user *filename, struct timeval __user *utimes)
+{
+	return sys_futimesat(AT_FDCWD, filename, utimes);
 }
 
 
@@ -476,7 +482,7 @@ asmlinkage long sys_utimes(char __user * filename, struct timeval __user * utime
  * We do this by temporarily clearing all FS-related capabilities and
  * switching the fsuid/fsgid around to the real ones.
  */
-asmlinkage long sys_access(const char __user * filename, int mode)
+asmlinkage long sys_faccessat(int dfd, const char __user *filename, int mode)
 {
 	struct nameidata nd;
 	int old_fsuid, old_fsgid;
@@ -506,7 +512,7 @@ asmlinkage long sys_access(const char __user * filename, int mode)
 	else
 		current->cap_effective = current->cap_permitted;
 
-	res = __user_walk(filename, LOOKUP_FOLLOW|LOOKUP_ACCESS, &nd);
+	res = __user_walk_fd(dfd, filename, LOOKUP_FOLLOW|LOOKUP_ACCESS, &nd);
 	if (!res) {
 		res = vfs_permission(&nd, mode);
 		/* SuS v2 requires we report a read only fs too */
@@ -523,6 +529,11 @@ asmlinkage long sys_access(const char __user * filename, int mode)
 	return res;
 }
 
+asmlinkage long sys_access(const char __user *filename, int mode)
+{
+	return sys_faccessat(AT_FDCWD, filename, mode);
+}
+
 asmlinkage long sys_chdir(const char __user * filename)
 {
 	struct nameidata nd;
@@ -635,14 +646,15 @@ out:
 	return err;
 }
 
-asmlinkage long sys_chmod(const char __user * filename, mode_t mode)
+asmlinkage long sys_fchmodat(int dfd, const char __user *filename,
+			     mode_t mode)
 {
 	struct nameidata nd;
 	struct inode * inode;
 	int error;
 	struct iattr newattrs;
 
-	error = user_path_walk(filename, &nd);
+	error = __user_walk_fd(dfd, filename, LOOKUP_FOLLOW, &nd);
 	if (error)
 		goto out;
 	inode = nd.dentry->d_inode;
@@ -669,6 +681,11 @@ out:
 	return error;
 }
 
+asmlinkage long sys_chmod(const char __user *filename, mode_t mode)
+{
+	return sys_fchmodat(AT_FDCWD, filename, mode);
+}
+
 static int chown_common(struct dentry * dentry, uid_t user, gid_t group)
 {
 	struct inode * inode;
@@ -717,6 +734,26 @@ asmlinkage long sys_chown(const char __user * filename, uid_t user, gid_t group)
 	return error;
 }
 
+asmlinkage long sys_fchownat(int dfd, const char __user *filename, uid_t user,
+			     gid_t group, int flag)
+{
+	struct nameidata nd;
+	int error = -EINVAL;
+	int follow;
+
+	if ((flag & ~AT_SYMLINK_NOFOLLOW) != 0)
+		goto out;
+
+	follow = (flag & AT_SYMLINK_NOFOLLOW) ? 0 : LOOKUP_FOLLOW;
+	error = __user_walk_fd(dfd, filename, follow, &nd);
+	if (!error) {
+		error = chown_common(nd.dentry, user, group);
+		path_release(&nd);
+	}
+out:
+	return error;
+}
+
 asmlinkage long sys_lchown(const char __user * filename, uid_t user, gid_t group)
 {
 	struct nameidata nd;
@@ -820,7 +857,8 @@ cleanup_file:
  * for the internal routines (ie open_namei()/follow_link() etc). 00 is
  * used by symlinks.
  */
-struct file *filp_open(const char * filename, int flags, int mode)
+static struct file *do_filp_open(int dfd, const char *filename, int flags,
+				 int mode)
 {
 	int namei_flags, error;
 	struct nameidata nd;
@@ -829,12 +867,17 @@ struct file *filp_open(const char * filename, int flags, int mode)
 	if ((namei_flags+1) & O_ACCMODE)
 		namei_flags++;
 
-	error = open_namei(filename, namei_flags, mode, &nd);
+	error = open_namei(dfd, filename, namei_flags, mode, &nd);
 	if (!error)
 		return nameidata_to_filp(&nd, flags);
 
 	return ERR_PTR(error);
 }
+
+struct file *filp_open(const char *filename, int flags, int mode)
+{
+	return do_filp_open(AT_FDCWD, filename, flags, mode);
+}
 EXPORT_SYMBOL(filp_open);
 
 /**
@@ -991,7 +1034,7 @@ void fastcall put_unused_fd(unsigned int fd)
 EXPORT_SYMBOL(put_unused_fd);
 
 /*
- * Install a file pointer in the fd array.  
+ * Install a file pointer in the fd array.
  *
  * The VFS is full of places where we drop the files lock between
  * setting the open_fds bitmap and installing the file in the file
@@ -1016,7 +1059,7 @@ void fastcall fd_install(unsigned int fd, struct file * file)
 
 EXPORT_SYMBOL(fd_install);
 
-long do_sys_open(const char __user *filename, int flags, int mode)
+long do_sys_open(int dfd, const char __user *filename, int flags, int mode)
 {
 	char *tmp = getname(filename);
 	int fd = PTR_ERR(tmp);
@@ -1024,7 +1067,7 @@ long do_sys_open(const char __user *filename, int flags, int mode)
 	if (!IS_ERR(tmp)) {
 		fd = get_unused_fd();
 		if (fd >= 0) {
-			struct file *f = filp_open(tmp, flags, mode);
+			struct file *f = do_filp_open(dfd, tmp, flags, mode);
 			if (IS_ERR(f)) {
 				put_unused_fd(fd);
 				fd = PTR_ERR(f);
@@ -1043,10 +1086,20 @@ asmlinkage long sys_open(const char __user *filename, int flags, int mode)
 	if (force_o_largefile())
 		flags |= O_LARGEFILE;
 
-	return do_sys_open(filename, flags, mode);
+	return do_sys_open(AT_FDCWD, filename, flags, mode);
 }
 EXPORT_SYMBOL_GPL(sys_open);
 
+asmlinkage long sys_openat(int dfd, const char __user *filename, int flags,
+			   int mode)
+{
+	if (force_o_largefile())
+		flags |= O_LARGEFILE;
+
+	return do_sys_open(dfd, filename, flags, mode);
+}
+EXPORT_SYMBOL_GPL(sys_openat);
+
 #ifndef __alpha__
 
 /*