Date:   Mon, 13 Feb 2023 08:30:44 +0300
From:   Dan Carpenter <error27@gmail.com>
To:     Guillaume Nault <gnault@redhat.com>
Cc:     oe-kbuild@lists.linux.dev, Cong Wang <cong.wang@bytedance.com>,
        lkp@intel.com, oe-kbuild-all@lists.linux.dev,
        linux-kernel@vger.kernel.org
Subject: Re: net/l2tp/l2tp_core.c:1481 l2tp_tunnel_register() warn: missing
 error code 'ret'
Message-ID: <Y+nLBPj70/03q8Do@kadam>
References: <202302031144.yY6UaRcD-lkp@intel.com>
 <Y9z8XPvcv5Wn2J8n@debian>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <Y9z8XPvcv5Wn2J8n@debian>
Precedence: bulk

On Fri, Feb 03, 2023 at 01:21:48PM +0100, Guillaume Nault wrote:
> On Fri, Feb 03, 2023 at 11:56:01AM +0300, Dan Carpenter wrote:
> > tree:   https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
> > head:   66a87fff1a87c260452f5a57123891ca5258c449
> > commit: c4d48a58f32c5972174a1d01c33b296fe378cce0 l2tp: convert l2tp_tunnel_list to idr
> > config: powerpc-randconfig-m031-20230202 (https://download.01.org/0day-ci/archive/20230203/202302031144.yY6UaRcD-lkp@intel.com/config)
> > compiler: powerpc-linux-gcc (GCC) 12.1.0
> > 
> > If you fix the issue, kindly add following tag where applicable
> > | Reported-by: kernel test robot <lkp@intel.com>
> > | Reported-by: Dan Carpenter <error27@gmail.com>
> > 
> > smatch warnings:
> > net/l2tp/l2tp_core.c:1481 l2tp_tunnel_register() warn: missing error code 'ret'
> > 
> > vim +/ret +1481 net/l2tp/l2tp_core.c
> > 
> > 6b9f34239b00e6 Guillaume Nault 2018-04-10  1456  int l2tp_tunnel_register(struct l2tp_tunnel *tunnel, struct net *net,
> > 6b9f34239b00e6 Guillaume Nault 2018-04-10  1457  			 struct l2tp_tunnel_cfg *cfg)
> > 6b9f34239b00e6 Guillaume Nault 2018-04-10  1458  {
> > c4d48a58f32c59 Cong Wang       2023-01-13  1459  	struct l2tp_net *pn = l2tp_pernet(net);
> > c4d48a58f32c59 Cong Wang       2023-01-13  1460  	u32 tunnel_id = tunnel->tunnel_id;
> > 6b9f34239b00e6 Guillaume Nault 2018-04-10  1461  	struct socket *sock;
> > 6b9f34239b00e6 Guillaume Nault 2018-04-10  1462  	struct sock *sk;
> > 6b9f34239b00e6 Guillaume Nault 2018-04-10  1463  	int ret;
> > 6b9f34239b00e6 Guillaume Nault 2018-04-10  1464  
> > c4d48a58f32c59 Cong Wang       2023-01-13  1465  	spin_lock_bh(&pn->l2tp_tunnel_idr_lock);
> > c4d48a58f32c59 Cong Wang       2023-01-13  1466  	ret = idr_alloc_u32(&pn->l2tp_tunnel_idr, NULL, &tunnel_id, tunnel_id,
> > c4d48a58f32c59 Cong Wang       2023-01-13  1467  			    GFP_ATOMIC);
> > c4d48a58f32c59 Cong Wang       2023-01-13  1468  	spin_unlock_bh(&pn->l2tp_tunnel_idr_lock);
> > c4d48a58f32c59 Cong Wang       2023-01-13  1469  	if (ret)
> > c4d48a58f32c59 Cong Wang       2023-01-13  1470  		return ret == -ENOSPC ? -EEXIST : ret;
> > c4d48a58f32c59 Cong Wang       2023-01-13  1471  
> > 6b9f34239b00e6 Guillaume Nault 2018-04-10  1472  	if (tunnel->fd < 0) {
> > 6b9f34239b00e6 Guillaume Nault 2018-04-10  1473  		ret = l2tp_tunnel_sock_create(net, tunnel->tunnel_id,
> > 6b9f34239b00e6 Guillaume Nault 2018-04-10  1474  					      tunnel->peer_tunnel_id, cfg,
> > 6b9f34239b00e6 Guillaume Nault 2018-04-10  1475  					      &sock);
> > 6b9f34239b00e6 Guillaume Nault 2018-04-10  1476  		if (ret < 0)
> > 6b9f34239b00e6 Guillaume Nault 2018-04-10  1477  			goto err;
> > 6b9f34239b00e6 Guillaume Nault 2018-04-10  1478  	} else {
> > 6b9f34239b00e6 Guillaume Nault 2018-04-10  1479  		sock = sockfd_lookup(tunnel->fd, &ret);
> > 6b9f34239b00e6 Guillaume Nault 2018-04-10  1480  		if (!sock)
> > 6b9f34239b00e6 Guillaume Nault 2018-04-10 @1481  			goto err;
> >                                                                         ^^^^^^^^^
> > I don't know why this is showing up as a 3 week old warning when it
> > looks like the code is from 2018...  Anyway, should this be an error
> > path or a success path?
> 
> This is an error path.
> But I don't understand this warning. Does it complain that 'ret' isn't
> initialised before the 'goto err;' jump? (this is done by
> sockfd_lookup() in case of error).

Or sorry, I didn't see the &ret.  Yes, Smatch thinks "ret" is zero here.
The kbuild-bot can't use the cross function database (building the DB
is too slow to scale).  So that's why the warning is printed.

regards,
dan carpenter