Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Mon, 13 May 2019 20:57:20 +1000
From:   "Tobin C. Harding" <me@tobin.cc>
To:     Nikolay Borisov <nborisov@suse.com>
Cc:     "Tobin C. Harding" <tobin@kernel.org>, Chris Mason <clm@fb.com>,
        Josef Bacik <josef@toxicpanda.com>,
        David Sterba <dsterba@suse.com>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        "Rafael J. Wysocki" <rafael@kernel.org>,
        linux-btrfs@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 2/2] fs: btrfs: Don't leak memory when failing add fsid
Message-ID: <20190513105720.GB15053@eros.localdomain>
References: <20190513033912.3436-1-tobin@kernel.org>
 <20190513033912.3436-3-tobin@kernel.org>
 <3473fcfa-88cf-b372-3beb-69d59320d50a@suse.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <3473fcfa-88cf-b372-3beb-69d59320d50a@suse.com>
User-Agent: Mutt/1.11.4 (2019-03-13)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Mon, May 13, 2019 at 09:04:49AM +0300, Nikolay Borisov wrote:
> 
> 
> On 13.05.19 г. 6:39 ч., Tobin C. Harding wrote:
> > A failed call to kobject_init_and_add() must be followed by a call to
> > kobject_put().  Currently in the error path when adding fs_devices we
> > are missing this call.  This could be fixed by calling
> > btrfs_sysfs_remove_fsid() if btrfs_sysfs_add_fsid() returns an error or
> > by adding a call to kobject_put() directly in btrfs_sysfs_add_fsid().
> > Here we choose the second option because it prevents the slightly
> > unusual error path handling requirements of kobject from leaking out
> > into btrfs functions.
> > 
> > Add a call to kobject_put() in the error path of kobject_add_and_init().
> > This causes the release method to be called if kobject_init_and_add()
> > fails.  open_tree() is the function that calls btrfs_sysfs_add_fsid()
> > and the error code in this function is already written with the
> > assumption that the release method is called during the error path of
> > open_tree() (as seen by the call to btrfs_sysfs_remove_fsid() under the
> > fail_fsdev_sysfs label).
> 
> I'm not familiar with the internals of kobject but
> btrfs_sysfs_remove_fsid calls __btrfs_sysfs_remove_fsid which in turn
> does kobject_del followed by kobject_put so its sequence is not exactly
> identical with your change. Presumably kobject_del is only required if
> you want to dispose of successfully registered sysfs node. This implies
> that __btrfs_sysfs_remove_fsid is actually broken when it comes to
> handling failed sysfs_add_fsid?

kobject_del() is not technically required in __btrfs_sysfs_remove_fsid()
since if kobject_put() drops the reference count to 0 and kobject_del()
has not been called then the kobject infrastructure will call
kobject_del() for us (and we get a pr_debug() message).  The code
sequence is correct although not _exactly_ written as the kobject
authors intended (I am not one of those authors, I'm just learning).

Thanks for looking at this.

	Tobin