Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   Saubhik Mukherjee <saubhik.mukherjee@gmail.com>
To:     isdn@linux-pingi.de, gregkh@linuxfoundation.org,
        jirislaby@kernel.org, dsterba@suse.com, jcmvbkbc@gmail.com,
        johannes@sipsolutions.net, kuba@kernel.org
Cc:     netdev@vger.kernel.org, linux-kernel@vger.kernel.org,
        andrianov@ispras.ru
Subject: [question] potential race between capinc_tty_init & capi_release
Message-ID: <af56f61a-6343-85fd-3efc-b3a2890246ac@gmail.com>
Date:   Tue, 17 Aug 2021 14:43:22 +0530
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101
 Thunderbird/78.13.0
MIME-Version: 1.0
Content-Type: text/plain; charset=windows-1252; format=flowed
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk

In drivers/isdn/capi/capi.c, based on the output of a static analysis 
tool, we found the possibility of the following race condition:

In capi_init, register_chrdev registers file operations callbacks, 
capi_fops. Then capinc_tty_init is executed.

Simultaneously the following chain of calls can occur (after a 
successful capi_open call).

capi_release -> capincci_free -> capincci_free_minor -> capiminor_free 
-> tty_unregister_device

tty_unregister_device reads capinc_tty_driver, which might not have been 
initialized at this point. So, we have a race between capi_release and 
capinc_tty_init.

If this is a possible race scenario, maybe moving register_chrdev after 
capinc_tty_init could fix it. But I am not sure if this will break 
something else. Please let me know if this is a potential race and can 
be fixed as mentioned.

Since this is based on a static analysis tool, this is not tested.