linux-kernel - Re: [PATCH] net: bluetooth: hci_sock: Fix a possible null-pointer dereference in hci_mgmt

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 for Android: free password hash cracker in your pocket

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

Message-Id: <BC138E23-E2FE-450E-B33E-1AD846D14687@holtmann.org>
Date:   Thu, 25 Jul 2019 11:39:32 +0200
From:   Marcel Holtmann <marcel@...tmann.org>
To:     Jia-Ju Bai <baijiaju1990@...il.com>
Cc:     Johan Hedberg <johan.hedberg@...il.com>,
        "David S. Miller" <davem@...emloft.net>,
        linux-bluetooth@...r.kernel.org, netdev@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] net: bluetooth: hci_sock: Fix a possible null-pointer
 dereference in hci_mgmt_cmd()

Hi Jia-Ju,

> In hci_mgmt_cmd(), there is an if statement on line 1570 to check
> whether hdev is NULL:
>    if (hdev && chan->hdev_init)
> 
> When hdev is NULL, it is used on line 1575:
>    err = handler->func(sk, hdev, cp, len);
> 
> Some called functions of handler->func use hdev, such as:
> set_appearance(), add_device() and remove_device() in mgmt.c.
> 
> Thus, a possible null-pointer dereference may occur.
> 
> To fix this bug, hdev is checked before calling handler->func().
> 
> This bug is found by a static analysis tool STCheck written by us.
> 
> Signed-off-by: Jia-Ju Bai <baijiaju1990@...il.com>
> ---
> net/bluetooth/hci_sock.c | 11 ++++++-----
> 1 file changed, 6 insertions(+), 5 deletions(-)
> 
> diff --git a/net/bluetooth/hci_sock.c b/net/bluetooth/hci_sock.c
> index d32077b28433..18ea1e47ea48 100644
> --- a/net/bluetooth/hci_sock.c
> +++ b/net/bluetooth/hci_sock.c
> @@ -1570,11 +1570,12 @@ static int hci_mgmt_cmd(struct hci_mgmt_chan *chan, struct sock *sk,
> 	if (hdev && chan->hdev_init)
> 		chan->hdev_init(sk, hdev);
> 
> -	cp = buf + sizeof(*hdr);
> -
> -	err = handler->func(sk, hdev, cp, len);
> -	if (err < 0)
> -		goto done;
> +	if (hdev) {
> +		cp = buf + sizeof(*hdr);
> +		err = handler->func(sk, hdev, cp, len);
> +		if (err < 0)
> +			goto done;
> +	}
> 
> 	err = msglen;

have you evaluated the statement above:

        no_hdev = (handler->flags & HCI_MGMT_NO_HDEV);                           
        if (no_hdev != !hdev) {                                                  
                err = mgmt_cmd_status(sk, index, opcode,                         
                                      MGMT_STATUS_INVALID_INDEX);                
                goto done;                                                       
        }

I think that code is just overly complex and can be simplified, but I doubt you get to the situation where hdev is NULL for any function that requires it. Only the handler->func marked with HCI_MGMT_NO_HDEV will get hdev == NULL and these are not using it.

So we might can make this easier code to really check the index != MGMT_INDEX_NONE check above to cover all cases to ensure that hdev is either valid or set to NULL before proceeding any further.

And since we have a full set of unit tests in tools/mgmt-tester, I assume we would have had a chance to catch an issue like this. But we can add a test case to it to explicitly call the functions with either MGMT_INDEX_NONE used or not.

Regards

Marcel