netdev - Re: crypto: deadlock between crypto_alg_sem/rtnl_mutex/genl

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: <20170315113019.GA28050@oracle.com>
Date:   Wed, 15 Mar 2017 07:30:19 -0400
From:   Sowmini Varadhan <sowmini.varadhan@...cle.com>
To:     Dmitry Vyukov <dvyukov@...gle.com>
Cc:     Herbert Xu <herbert@...dor.apana.org.au>,
        David Miller <davem@...emloft.net>,
        linux-crypto@...r.kernel.org, LKML <linux-kernel@...r.kernel.org>,
        Eric Dumazet <edumazet@...gle.com>,
        Cong Wang <xiyou.wangcong@...il.com>,
        netdev <netdev@...r.kernel.org>, santosh.shilimkar@...cle.com,
        rds-devel@....oracle.com, syzkaller <syzkaller@...glegroups.com>
Subject: Re: crypto: deadlock between crypto_alg_sem/rtnl_mutex/genl_mutex

On (03/15/17 10:08), Dmitry Vyukov wrote:
> After I've applied the patch these reports stopped to happen, and I
> have not seem any other reports that look relevant.
> However, it there was one, but it looks like a different issue and it
> was probably masked by massive amounts of original deadlock reports:

Yes, this looks like a valid deadlock.

I think there may be some ->dumpit callbacks that take the rtnl_lock
and do not unlock it before return, e.g., I see nl80211_dump_interface()
doing this at 

   2612         rtnl_lock();
   2613         if (!cb->args[2]) {
     :
   2619                 ret = nl80211_dump_wiphy_parse(skb, cb, &state);
   2620                 if (ret)
   2621                         return ret;

afaict, nl80211_dump_wiphy_parse does not itself do rtnl_unlock on error.


If that's the case then we'd run into the circular locking dependancy
flagged by lockdep. 

Disclaimer: I did not check every single ->dumpit, there may be more
than one of these..