linux-kernel - Re: [PATCH 5.10 09/38] llc: fix netdevice reference leaks in llc_ui

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] [day] [month] [year] [list]

Message-ID: <YkGDEzS7bWHNIoP3@kroah.com>
Date:   Mon, 28 Mar 2022 11:42:43 +0200
From:   Greg Kroah-Hartman <gregkh@...uxfoundation.org>
To:     Pavel Machek <pavel@...x.de>
Cc:     Jakub Kicinski <kuba@...nel.org>, linux-kernel@...r.kernel.org,
        stable@...r.kernel.org, Eric Dumazet <edumazet@...gle.com>,
        赵子轩 <beraphin@...il.com>,
        Stoyan Manolov <smanolov@...e.de>
Subject: Re: [PATCH 5.10 09/38] llc: fix netdevice reference leaks in
 llc_ui_bind()

On Mon, Mar 28, 2022 at 11:31:16AM +0200, Pavel Machek wrote:
> Hi!
> 
> > > > > Should be in mainline on Thursday, LMK if we need to accelerate.
> > > > > IDK if anyone enables LLC2.
> > > > 
> > > > I'll queue this up now, thanks.
> > > 
> > > As the changelog says, this needs b37a46683739, otherwise there will
> > > be oops-es in even more cases.
> > 
> > If you look at the change, I think I already handled that issue.  If
> > not, please let me know.
> 
> I did not notice you making changes there, but no, it is not correct
> AFAICT.
> 
> # commit 163960a7de1333514c9352deb7c80c6b9fd9abf2
> # Author: Eric Dumazet <edumazet@...gle.com>
> # Date:   Thu Mar 24 20:58:27 2022 -0700
> 
> #    llc: only change llc->dev when bind() succeeds
> ...    
> #     Make sure commit b37a46683739 ("netdevice: add the case if dev is NULL")
> #     is already present in your trees.
> 
> Before b37a46683739, dev_put can't handle NULL.
>     
> +++ b/net/llc/af_llc.c
> @@ -287,14 +288,14 @@ static int llc_ui_autobind(struct socket *sock, struct sockaddr_llc *addr)
> ...
> 
> -		llc->dev = dev_getfirstbyhwtype(&init_net, addr->sllc_arphrd);
> -	if (!llc->dev)
> +		dev = dev_getfirstbyhwtype(&init_net, addr->sllc_arphrd);
> +	if (!dev)
>  		goto out;
>  	rc = -EUSERS;
>  	llc->laddr.lsap = llc_ui_autoport();
> 
> One of several paths where we goto out with dev==NULL.
> 
> @@ -311,10 +317,7 @@ static int llc_ui_autobind(struct socket *sock, struct sockaddr_llc *addr)
>  	sock_reset_flag(sk, SOCK_ZAPPED);
>  	rc = 0;
>  out:
> -	if (rc) {
> -		dev_put(llc->dev);
> -		llc->dev = NULL;
> -	}
> +	dev_put(dev);
>  	return rc;
>  }
> 
> 
> But dev_put can't handle NULL.

Ah, missed that one.  I'll go queue up b37a46683739 now.

thanks,

greg k-h