Re: [PATCH v3] fs: introduce getfsxattrat and setfsxattrat syscalls

On Tue 11-02-25 18:22:47, Andrey Albershteyn wrote:
> From: Andrey Albershteyn <[email protected]>
> 
> Introduce getfsxattrat and setfsxattrat syscalls to manipulate inode
> extended attributes/flags. The syscalls take parent directory fd and
> path to the child together with struct fsxattr.
> 
> This is an alternative to FS_IOC_FSSETXATTR ioctl with a difference
> that file don't need to be open as we can reference it with a path
> instead of fd. By having this we can manipulated inode extended
> attributes not only on regular files but also on special ones. This
> is not possible with FS_IOC_FSSETXATTR ioctl as with special files
> we can not call ioctl() directly on the filesystem inode using fd.
> 
> This patch adds two new syscalls which allows userspace to get/set
> extended inode attributes on special files by using parent directory
> and a path - *at() like syscall.
> 
> Also, as vfs_fileattr_set() is now will be called on special files
> too, let's forbid any other attributes except projid and nextents
> (symlink can have an extent).
> 
> CC: [email protected]
> CC: [email protected]
> CC: [email protected]
> Signed-off-by: Andrey Albershteyn <[email protected]>

Some comments below:

> +SYSCALL_DEFINE4(getfsxattrat, int, dfd, const char __user *, filename,
> +		struct fsxattr __user *, fsx, unsigned int, at_flags)
> +{
> +	CLASS(fd, dir)(dfd);
> +	struct fileattr fa;
> +	struct path filepath;
> +	int error;
> +	unsigned int lookup_flags = 0;
> +
> +	if ((at_flags & ~(AT_SYMLINK_NOFOLLOW | AT_EMPTY_PATH)) != 0)
> +		return -EINVAL;
> +
> +	if (at_flags & AT_SYMLINK_FOLLOW)
	    ^^ This should be !(at_flags & AT_SYMLINK_NOFOLLOW)?

In the check above you verify for AT_SYMLINK_NOFOLLOW and that also matches
what setxattrat() does...


> +		lookup_flags |= LOOKUP_FOLLOW;
> +
> +	if (at_flags & AT_EMPTY_PATH)
> +		lookup_flags |= LOOKUP_EMPTY;
> +
> +	if (fd_empty(dir))
> +		return -EBADF;

This check is wrong and in fact the whole dfd handling looks buggy.
openat(2) manpage describes the expected behavior:

       The dirfd argument is used in conjunction with the pathname argument as
       follows:

       •  If the pathname given in pathname is absolute,  then  dirfd  is  ig-
          nored.
	  ^^^^ This is what you break. If the pathname is absolute, you're
not expected to touch dirfd.

       •  If  the pathname given in pathname is relative and dirfd is the spe-
          cial value AT_FDCWD, then pathname is interpreted  relative  to  the
          current working directory of the calling process (like open()).
          ^^^ Also AT_FDCWD handling would be broken by the above check.

       •  If  the  pathname  given  in pathname is relative, then it is inter-
          preted relative to the directory referred to by the file  descriptor
          dirfd  (rather than relative to the current working directory of the
          calling process, as is done by open() for a relative pathname).   In
          this  case,  dirfd  must  be a directory that was opened for reading
          (O_RDONLY) or using the O_PATH flag.

       If the pathname given in pathname is relative, and dirfd is not a valid
       file descriptor, an error (EBADF) results.  (Specifying an invalid file
       descriptor number in dirfd can be used as a means to ensure that  path-
       name is absolute.)

> +
> +	error = user_path_at(dfd, filename, lookup_flags, &filepath);
		^^^ And user_path_at() isn't quite what you need either
because with AT_EMPTY_PATH we also want to allow for filename to be NULL
(not just empty string) and user_path_at() does not support that. That's
why I in my previous replies suggested you should follow what setxattrat()
does and that sadly it is more painful than it should be. You need
something like:

	name = getname_maybe_null(filename, at_flags);
	if (!name) {
		CLASS(fd, f)(dfd);

		if (fd_empty(f))
			return -EBADF;
		error = vfs_fileattr_get(file_dentry(fd_file(f)), &fa);
	} else {
		error = filename_lookup(dfd, filename, lookup_flags, &filepath,
					NULL);
		if (error)
			goto out;
		error = vfs_fileattr_get(filepath.dentry, &fa);
		path_put(&filepath);
	}
	if (!error)
		error = copy_fsxattr_to_user(&fa, fsx);
out:
	putname(name);
	return error;

Longer term, we need to provide user_path_maybe_null_at() for this but I
don't want to drag you into this cleanup :)

> +	if (error)
> +		return error;
> +
> +	error = vfs_fileattr_get(filepath.dentry, &fa);
> +	if (!error)
> +		error = copy_fsxattr_to_user(&fa, fsx);
> +
> +	path_put(&filepath);
> +	return error;
> +}
> +
> +SYSCALL_DEFINE4(setfsxattrat, int, dfd, const char __user *, filename,
> +		struct fsxattr __user *, fsx, unsigned int, at_flags)
> +{
> +	CLASS(fd, dir)(dfd);
> +	struct fileattr fa;
> +	struct path filepath;
> +	int error;
> +	unsigned int lookup_flags = 0;
> +
> +	if ((at_flags & ~(AT_SYMLINK_FOLLOW | AT_EMPTY_PATH)) != 0)
> +		return -EINVAL;
> +
> +	if (at_flags & AT_SYMLINK_FOLLOW)
> +		lookup_flags |= LOOKUP_FOLLOW;

I think using AT_SYMLINK_NOFOLLOW is actually more traditional and thus
less surprising to users so I'd prefer that. Definitely this needs to be
consistent with getfsxattrat().

> +
> +	if (at_flags & AT_EMPTY_PATH)
> +		lookup_flags |= LOOKUP_EMPTY;
> +
> +	if (fd_empty(dir))
> +		return -EBADF;

Same comment regarding dfd handling as above.

> +
> +	if (copy_fsxattr_from_user(&fa, fsx))
> +		return -EFAULT;
> +
> +	error = user_path_at(dfd, filename, lookup_flags, &filepath);
> +	if (error)
> +		return error;
> +
> +	error = mnt_want_write(filepath.mnt);
> +	if (!error) {
> +		error = vfs_fileattr_set(file_mnt_idmap(fd_file(dir)),
> +					 filepath.dentry, &fa);
> +		mnt_drop_write(filepath.mnt);
> +	}
> +
> +	path_put(&filepath);
> +	return error;
> +}

Otherwise the patch looks good to me.

								Honza
-- 
Jan Kara <[email protected]>
SUSE Labs, CR