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]
Message-ID: <Zys6KGmEWVnwidLb@pengutronix.de>
Date: Wed, 6 Nov 2024 10:43:04 +0100
From: Oleksij Rempel <o.rempel@...gutronix.de>
To: Jiri Pirko <jiri@...nulli.us>
Cc: Dmitry Antipov <dmantipov@...dex.ru>,
	Robin van der Gracht <robin@...tonic.nl>,
	Oliver Hartkopp <socketcan@...tkopp.net>,
	Marc Kleine-Budde <mkl@...gutronix.de>, linux-can@...r.kernel.org,
	netdev@...r.kernel.org, lvc-project@...uxtesting.org,
	syzbot+d4e8dc385d9258220c31@...kaller.appspotmail.com
Subject: Re: [PATCH v2] can: fix skb reference counting in j1939_session_new()

On Tue, Nov 05, 2024 at 05:37:53PM +0100, Jiri Pirko wrote:
> Tue, Nov 05, 2024 at 10:48:23AM CET, dmantipov@...dex.ru wrote:
> >Since 'j1939_session_skb_queue()' do an extra 'skb_get()' for each
> >new skb, I assume that the same should be done for an initial one
> 
> It is odd to write "I assume" for fix like this. You should know for
> sure, don't you?

Hm... looks the there is more then one refcounting problem at this
point. skb_queue is set from 3 different paths, with resulting 3 different
refcount states:

j1939_sk_send_loop()
  skb = j1939_sk_alloc_skb() // skb with refcount == 1
  if (!session) {
    session = j1939_tp_send(priv, skb, size)
       ... 
       session = j1939_session_new(priv, skb, size);
          skb_queue_tail(&session->skb_queue, skb); // skb refcount == 1
          
  } else {
    j1939_session_skb_queue(session, skb);
      // here, skb is refcounted
      skb_queue_tail(&session->skb_queue, skb_get(skb)); // skb refcount == 2
  }
  
  // at the end of function, skb refcount == 1 or 2
     
j1939_xtp_rx_rts_session_new()
  j1939_session_fresh_new()
    skb = alloc_skb() // skb with refcount == 1
    session = j1939_session_new(priv, skb, size);
       skb_queue_tail(&session->skb_queue, skb);
    skb_put(skb, size); // skb with refcount == 0

I agree with this patch, but there is missing skb_put() in j1939_sk_send_loop()

> 
> >in 'j1939_session_new()' just to avoid refcount underflow.
> >
> >Reported-by: syzbot+d4e8dc385d9258220c31@...kaller.appspotmail.com
> >Closes: https://syzkaller.appspot.com/bug?extid=d4e8dc385d9258220c31
> >Fixes: 9d71dd0c7009 ("can: add support of SAE J1939 protocol")
> >Signed-off-by: Dmitry Antipov <dmantipov@...dex.ru>
> >---
> >v2: resend after hitting skb refcount underflow once again when looking
> >around https://syzkaller.appspot.com/bug?extid=0e6ddb1ef80986bdfe64
> >---
> > net/can/j1939/transport.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> >diff --git a/net/can/j1939/transport.c b/net/can/j1939/transport.c
> >index 319f47df3330..95f7a7e65a73 100644
> >--- a/net/can/j1939/transport.c
> >+++ b/net/can/j1939/transport.c
> >@@ -1505,7 +1505,7 @@ static struct j1939_session *j1939_session_new(struct j1939_priv *priv,
> > 	session->state = J1939_SESSION_NEW;
> > 
> > 	skb_queue_head_init(&session->skb_queue);
> >-	skb_queue_tail(&session->skb_queue, skb);
> >+	skb_queue_tail(&session->skb_queue, skb_get(skb));
> > 
> > 	skcb = j1939_skb_to_cb(skb);
> > 	memcpy(&session->skcb, skcb, sizeof(session->skcb));
> >-- 
> >2.47.0
> >
> >
> 

-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ