netdev - Re: ipmr, ip6mr: Unite dumproute flows

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]

Date:   Wed, 17 Oct 2018 21:49:30 +0300
From:   Nikolay Aleksandrov <nikolay@...ulusnetworks.com>
To:     Colin Ian King <colin.king@...onical.com>,
        Yuval Mintz <yuvalm@...lanox.com>
Cc:     "David S. Miller" <davem@...emloft.net>,
        Alexey Kuznetsov <kuznet@....inr.ac.ru>,
        Hideaki YOSHIFUJI <yoshfuji@...ux-ipv6.org>,
        "netdev@...r.kernel.org" <netdev@...r.kernel.org>
Subject: Re: ipmr, ip6mr: Unite dumproute flows

On 17/10/2018 21:22, Colin Ian King wrote:
> Hi,
> 
> Static analysis on linux-next has picked up a potential bug with commit:
> 
> commit 7b0db85737db3f4d76b2a412e4f19eae59b8b494
> Author: Yuval Mintz <yuvalm@...lanox.com>
> Date:   Wed Feb 28 23:29:39 2018 +0200
> 
>     ipmr, ip6mr: Unite dumproute flows
> 
> in function mr_table_dump(), s_e is and never seems to change, so the
> check if (e < s_e) is never true:  See code below, as annotated by
> CoverityScan:
> 
> 317        }
>    assignment: Assigning: e = 0U.
> 
> 318        e = 0;
>    assignment: Assigning: s_e = 0U.
> 
> 319        s_e = 0;
> 320
> 321        spin_lock_bh(lock);
> 322        list_for_each_entry(mfc, &mrt->mfc_unres_queue, list) {
>    at_least: At condition e < s_e, the value of e must be at least 0.
>    const: At condition e < s_e, the value of s_e must be equal to 0.
>    dead_error_condition: The condition e < s_e cannot be true.
> 
> 323                if (e < s_e)
> 
>    CID 1474511 (#1 of 1): Logically dead code (DEADCODE)
> dead_error_line: Execution cannot reach this statement: goto next_entry2;.
> 
> 324                        goto next_entry2;
> 325                if (filter->dev &&
> 326                    !mr_mfc_uses_dev(mrt, mfc, filter->dev))
> 327                        goto next_entry2;
> 328
> 329                err = fill(mrt, skb, NETLINK_CB(cb->skb).portid,
> 330                           cb->nlh->nlmsg_seq, mfc, RTM_NEWROUTE, flags);
> 331                if (err < 0) {
> 332                        spin_unlock_bh(lock);
> 333                        goto out;
> 334                }
> 335next_entry2:
> 
>    incr: Incrementing e. The value of e is now 1.
>    incr: Incrementing e. The value of e is now 2.
> 
> 336                e++;
> 
> 
> Note that the zero'ing of e and s_e for ipmr and ip6mr was added in by
> earlier commits:
> 
> commit 8fb472c09b9df478a062eacc7841448e40fc3c17
> Author: Nikolay Aleksandrov <nikolay@...ulusnetworks.com>
> Date:   Thu Jan 12 15:53:33 2017 +0100
> 
>     ipmr: improve hash scalability
> 
> commit 87c418bf1323d57140f4b448715f64de3fbb7e91
> Author: Yuval Mintz <yuvalm@...lanox.com>
> Date:   Wed Feb 28 23:29:31 2018 +0200
> 
>     ip6mr: Align hash implementation to ipmr
> 
> Anyhow, this looks incorrect, but I'm not familiar with this code to
> suggest the correct fix.
> 
> Colin
> 

Indeed, there's a case where we might miss an unresolved entry due to the zeroing.
I'll send a fix shortly after testing.

Thanks,
 Nik