Re: [RFC][PATCH 14/27] tricky: elevate write count files are open()ed

[Herbert Poetzl <herbert@13thfloor.at>]

On Wed, Jun 07, 2006 at 05:10:25PM -0700, Dave Hansen wrote:
> 
> This is the first really tricky patch in the series.  It
> elevates the writer count on a mount each time a
> non-special file is opened for write.
> 
> This is not completely apparent in the patch because the
> two if() conditions in may_open() above the
> mnt_want_write() call are, combined, equivalent to
> special_file().
> 
> There is also an elevated count around the vfs_create()
> call in open_namei().  The count does not need to be
> kept elevated all the way into the may_open() call because
> after creation, the write bits of the acc_mode are cleared.
> This keeps may_open() from ever failing.  Howver, this may
> open one potential race where a change from a r/w to a r/o
> mount could occur between the mnt_drop_write() and may_open()
> allowing a user to obtain a r/w file on what is now a r/w

probably means 'read only mount'

what about using atomic_dec_and_test() to avoid
having such cases (or a lock if required)?

best,
Herbert

> mount.  But, this functionality does not yet exist.
> 
> 
> Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
> ---
> 
>  lxc-dave/fs/file_table.c |    5 ++++-
>  lxc-dave/fs/namei.c      |   29 ++++++++++++++++++++++++++---
>  lxc-dave/ipc/mqueue.c    |    3 +++
>  3 files changed, 33 insertions(+), 4 deletions(-)
> 
> diff -puN fs/namei.c~elevate-writers-opens-part1 fs/namei.c
> --- lxc/fs/namei.c~elevate-writers-opens-part1	2006-06-07 16:53:20.000000000 -0700
> +++ lxc-dave/fs/namei.c	2006-06-07 16:53:20.000000000 -0700
> @@ -1511,8 +1511,17 @@ int may_open(struct nameidata *nd, int a
>  			return -EACCES;
>  
>  		flag &= ~O_TRUNC;
> -	} else if (IS_RDONLY(inode) && (flag & FMODE_WRITE))
> -		return -EROFS;
> +	} else if (flag & FMODE_WRITE) {
> +		/*
> +		 * effectively: !special_file()
> +		 * balanced by __fput()
> +		 */
> +		error = mnt_want_write(nd->mnt);
> +		if (error)
> +			return error;
> +		if (IS_RDONLY(inode))
> +			return -EROFS;
> +	}
>  	/*
>  	 * An append-only file must be opened in append mode for writing.
>  	 */
> @@ -1642,10 +1651,24 @@ do_last:
>  	if (!path.dentry->d_inode) {
>  		if (!IS_POSIXACL(dir->d_inode))
>  			mode &= ~current->fs->umask;
> -		error = vfs_create(dir->d_inode, path.dentry, mode, nd);
> +		/*
> +		 * this serves dual roles, it makes sure there is no
> +		 * r/o mount, and keeps the write count for what is
> +		 * the newly created file
> +		 */
> +		error = mnt_want_write(nd->mnt);
> +		if (!error)
> +			error = vfs_create(dir->d_inode, path.dentry, mode, nd);
>  		mutex_unlock(&dir->d_inode->i_mutex);
>  		dput(nd->dentry);
>  		nd->dentry = path.dentry;
> +		/*
> +		 * Unconditionally drop the write access because
> +		 * the acc_mode=0 set below will keep may_open()
> +		 * from ever failing if there was a r/o mount
> +		 * between here and there
> +		 */
> +		mnt_drop_write(nd->mnt);
>  		if (error)
>  			goto exit;
>  		/* Don't check for write permission, don't truncate */
> diff -puN fs/open.c~elevate-writers-opens-part1 fs/open.c
> diff -puN include/linux/mount.h~elevate-writers-opens-part1 include/linux/mount.h
> diff -puN fs/file_table.c~elevate-writers-opens-part1 fs/file_table.c
> --- lxc/fs/file_table.c~elevate-writers-opens-part1	2006-06-07 16:53:20.000000000 -0700
> +++ lxc-dave/fs/file_table.c	2006-06-07 16:53:20.000000000 -0700
> @@ -180,8 +180,11 @@ void fastcall __fput(struct file *file)
>  	if (unlikely(inode->i_cdev != NULL))
>  		cdev_put(inode->i_cdev);
>  	fops_put(file->f_op);
> -	if (file->f_mode & FMODE_WRITE)
> +	if (file->f_mode & FMODE_WRITE) {
>  		put_write_access(inode);
> +		if(!special_file(inode->i_mode))
> +			mnt_drop_write(mnt);
> +	}
>  	file_kill(file);
>  	file->f_dentry = NULL;
>  	file->f_vfsmnt = NULL;
> diff -puN ipc/mqueue.c~elevate-writers-opens-part1 ipc/mqueue.c
> --- lxc/ipc/mqueue.c~elevate-writers-opens-part1	2006-06-07 16:53:20.000000000 -0700
> +++ lxc-dave/ipc/mqueue.c	2006-06-07 16:53:20.000000000 -0700
> @@ -679,6 +679,9 @@ asmlinkage long sys_mq_open(const char _
>  				goto out;
>  			filp = do_open(dentry, oflag);
>  		} else {
> +			error = mnt_want_write(mqueue_mnt);
> +			if (error)
> +				goto out;
>  			filp = do_create(mqueue_mnt->mnt_root, dentry,
>  						oflag, mode, u_attr);
>  		}
> _