Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Sun, 13 Nov 2022 00:47:40 -0800
From:   Hyunwoo Kim <imv4bel@gmail.com>
To:     Eli Billauer <eli.billauer@gmail.com>
Cc:     gregkh@linuxfoundation.org, arnd@arndb.de,
        linux-kernel@vger.kernel.org, linux-usb@vger.kernel.org,
        stern@rowland.harvard.edu, imv4bel@gmail.com
Subject: Re: [PATCH v2] char: xillybus: Prevent use-after-free due to race
 condition
Message-ID: <20221113084740.GA6458@ubuntu>
References: <20221030094209.65916-1-eli.billauer@gmail.com>
 <20221113080558.GA5854@ubuntu>
 <2a8f59ac-9d49-ffa3-b035-809f2fac38ec@gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <2a8f59ac-9d49-ffa3-b035-809f2fac38ec@gmail.com>
Precedence: bulk

Dear,

On Sun, Nov 13, 2022 at 10:30:16AM +0200, Eli Billauer wrote:
> On 13/11/2022 10:05, Hyunwoo Kim wrote:
> > Dear,
> > 
> > Sorry for the late review.
> > 
> > This patch cannot prevent the UAF scenario I presented:
> > ```
> >                  cpu0                                                cpu1
> >         1. xillyusb_open()
> >            mutex_lock(&kref_mutex);    // unaffected lock
> >            xillybus_find_inode()
> >            mutex_lock(&unit_mutex);
> >            unit = iter;
> >            mutex_unlock(&unit_mutex);
> >                                                               2. xillyusb_disconnect()
> >                                                                  xillybus_cleanup_chrdev()
> >                                                                  mutex_lock(&unit_mutex);
> >                                                                  kfree(unit);
> >                                                                  mutex_unlock(&unit_mutex);
> >         3. *private_data = unit->private_data;   // UAF
> > 
> > ```
> > 
> > This is a UAF for 'unit', not a UAF for 'xdev'.
> > So, the added 'kref_mutex' has no effect.
> > 
> 
> You're correct. This submitted patch solves only one problem out of two. It
> prevents the content of @private_data to be freed, but you're right that
> @unit itself isn't protected well enough.
> 
> The problem you're pointing at is that @unit can be freed before its
> attempted use, because the mutex is released too early.
> 
> This is easily solved by moving down the mutex_unlock() call to after @unit
> has been used. Do you want the pleasure to submit this patch, or should I?

I hope you work as before.

And, even if the mutex_unlock(&unit_mutex); of xillybus_find_inode() 
is finally moved, xdev may be released before kref_get() is executed 
if xillyusb_disconnect() ends just before the function returns. 
(Of course, this is an extremely rare case.)

So, in xillyusb_open() we need to move kref_get() above xillybus_find_inode().


Regards,
Hyunwoo Kim