netdev - Re: 回覆: [PATCH net-next] net: ftgmac100: Fix potential NULL dereference in error handling

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

Message-ID: <a2dba28a-6ac4-4770-b618-acfdd59cbbf4@stanley.mountain>
Date: Tue, 10 Sep 2024 08:37:21 +0300
From: Dan Carpenter <dan.carpenter@...aro.org>
To: Andrew Lunn <andrew@...n.ch>
Cc: Jacky Chou <jacky_chou@...eedtech.com>,
	"David S. Miller" <davem@...emloft.net>,
	Eric Dumazet <edumazet@...gle.com>,
	Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>,
	Uwe Kleine-König <u.kleine-koenig@...gutronix.de>,
	Jacob Keller <jacob.e.keller@...el.com>,
	"netdev@...r.kernel.org" <netdev@...r.kernel.org>,
	"kernel-janitors@...r.kernel.org" <kernel-janitors@...r.kernel.org>
Subject: Re: 回覆: [PATCH net-next] net: ftgmac100: Fix potential NULL dereference in error handling

On Mon, Sep 09, 2024 at 02:03:32PM +0200, Andrew Lunn wrote:
> > > Are you actually saying:
> > > 
> > >         if (netdev->phydev) {
> > >                 /* If we have a PHY, start polling */
> > >                 phy_start(netdev->phydev);
> > >         }
> > > 
> > > is wrong, it is guaranteed there is always a phydev?
> > > 
> > This patch is focus on error handling when using NC-SI at open stage.
> > 
> >          if (netdev->phydev) {
> >                  /* If we have a PHY, start polling */
> >                  phy_start(netdev->phydev);
> >          }
> > 
> > This code is used to check the other cases.
> > Perhaps, phy-handle or fixed-link property are not added in DTS.
> 
> I'm guessing, but i think the static analysers see this condition, and
> deducing that phydev might be a NULL. Hence when phy_stop() is called,
> it needs the check.
> 
> You say the static analyser is wrong, probably because it cannot check
> the bigger context. It can be NULL for phy_start() but not for
> phy_stop(). Maybe you can give it some more hints?
> 
> Dan, is this Smatch? Is it possible to dump the paths through the code
> where it thinks it might be NULL?

Adding a check here is the correct thing.  The current code works because we
only have the one goto after the call to phy_start(netdev->phydev), but as
soon as we add a second goto then it will crash.

Silencing this warning means tying the information from probe() into it.  It's
a fun problem but not something I'm going to do this year.

regards,
dan carpenter