linux-kernel - Re: [PATCH 0/1] devpts: use dynamic

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Wed, 16 Aug 2017 17:46:46 -0500
From:   ebiederm@...ssion.com (Eric W. Biederman)
To:     Linus Torvalds <torvalds@...ux-foundation.org>
Cc:     Christian Brauner <christian.brauner@...onical.com>,
        Christian Brauner <christian.brauner@...ntu.com>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        "Serge E. Hallyn" <serge@...lyn.com>,
        Al Viro <viro@...iv.linux.org.uk>
Subject: Re: [PATCH 0/1] devpts: use dynamic_dname() to generate proc name

Linus Torvalds <torvalds@...ux-foundation.org> writes:

> On Wed, Aug 16, 2017 at 1:30 PM, Linus Torvalds
> <torvalds@...ux-foundation.org> wrote:
>>
>> I suspect the easiest fix is to just add a "mnt" argument to
>> devpts_acquire(),  It shouldn't be too painful. Let me try.
>
> Ok, here's a *very* lightly tested patch. It might have new bugs, but
> it makes your test program DTRT.
>
> Al, mind going over this and making sure I didn't miss anything?
>
> And Christian, if you can beat on this, that would be good.

Linus reading through this it looks like your error handling is wrong.

> diff --git a/drivers/tty/pty.c b/drivers/tty/pty.c
> index 284749fb0f6b..432f514e3f42 100644
> --- a/drivers/tty/pty.c
> +++ b/drivers/tty/pty.c
> @@ -793,6 +793,7 @@ static int ptmx_open(struct inode *inode, struct file *filp)
>  	struct tty_struct *tty;
>  	struct path *pts_path;
>  	struct dentry *dentry;
> +	struct vfsmount *mnt;
>  	int retval;
>  	int index;
>  
> @@ -805,7 +806,7 @@ static int ptmx_open(struct inode *inode, struct file *filp)
>  	if (retval)
>  		return retval;
>  
> -	fsi = devpts_acquire(filp);
> +	fsi = devpts_acquire(filp, &mnt);
>  	if (IS_ERR(fsi)) {
>  		retval = PTR_ERR(fsi);
>  		goto out_free_file;
> @@ -849,9 +850,14 @@ static int ptmx_open(struct inode *inode, struct file *filp)
>  	pts_path = kmalloc(sizeof(struct path), GFP_KERNEL);
>  	if (!pts_path)
>  		goto err_release;
> -	pts_path->mnt = filp->f_path.mnt;
> -	pts_path->dentry = dentry;
> -	path_get(pts_path);
> +
> +	/*
> +	 * The mnt already got a ref from devpts_acquire(),
> +	 * so we only dget() on the dentry.
> +	 */
> +	pts_path->mnt = mnt;
> +	pts_path->dentry = dget(dentry);
> +
>  	tty->link->driver_data = pts_path;
>  
>  	retval = ptm_driver->ops->open(tty, filp);
        ^^^^^^^

If this open fails the code jumps to err_put_path which falls
through into out_put_fsi.  But it also does path_put(pts_path).
Which will result in a double mntput of mnt.

So I think err_path_put needs to be updated to just put the dentry,
and let the later mntput put the mount.

> @@ -874,6 +880,7 @@ static int ptmx_open(struct inode *inode, struct file *filp)
>  	devpts_kill_index(fsi, index);
>  out_put_fsi:
>  	devpts_release(fsi);
> +	mntput(mnt);
>  out_free_file:
>  	tty_free_file(filp);
>  	return retval;

Eric